AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

HQ-CLIP: Leveraging Large Vision-Language Models to Create High-Quality Image-Text Datasets and CLIP Models

Zhixiang Wei, Guangting Wang, Xiaoxiao Ma, Ke Mei, Huaian Chen, Yi Jin, Fengyun Rao

ICCV 2025 · Vision-language and multimodal › CLIP and zero-shot · approach: CLIP training recipes and data curation · 6 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 on VLM-150M recaptioned data (mixed training) (ablation)DFN CLIP baseline · DataComp Mediumavg over 38 datasets: 36.8 -> 39.9Improved
add hard negative identification loss (ablation)mixed-training CLIP · DataComp Mediumavg over 38 datasets: 39.9 -> 40.7Improved
add short-tag classification loss (ablation)CLIP with HNI · DataComp Mediumavg over 38 datasets: 40.7 -> 41.1Improved
train HQ-CLIP instead of plain training on VLM-150M (ablation)VLM-150M CLIP · DataComp Largeavg over 38 datasets: 56.6 -> 58.6Improved
train on VLM-150M with plain CLIP lossDFN CLIP baseline · DataComp LargeImageNet: 68.7 -> 67.7Hurt
use 100% enriched text instead of 75% mix with raw captions (ablation)mixed training · DataComp Mediumavg over 38 datasets: 39.9 -> 36.9Hurt
mix raw captions with enriched text (75% enriched vs 0%) (ablation)DFN CLIP baseline · DataComp Mediumavg over 38 datasets: 36.8 -> 39.9Improved
increase hard negatives per image from 1 to 4 (ablation)HQ-CLIP · DataComp Mediumavg over 38 datasets: 40.1 -> 39.9No clear effect
vary STC tag vocabulary size from 3000 to 90000 (ablation)HQ-CLIP · DataComp Mediumavg over 38 datasets: 39.9 -> 40.5No clear effect
add caption input to Qwen2-VL-7B SFT refinement (ablation)Qwen2-VL-7B captioner · DataComp Mediumavg over 38 datasets: 38.7 -> 39.9Improved
scale captioner from 7B (SFT) to 72B (ablation)Qwen2-VL captioner · DataComp Mediumavg over 38 datasets: 39.9 -> 40.1No clear effect
use HQ-CLIP vision encoder instead of DFN in LLaVA-1.5LLaVA-1.5 ViT-B · LLaVA-1.5MME: 1452.1 -> 1574.0Improved

Later papers that compared with this method

1 later accepted paper reports HQ-CLIP as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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