Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
③[ICML2024] [IQA, IAA, VQA] All-in-one Foundation Model for visual scoring. Can efficiently fine-tune to downstream datasets.
| Date | Stars |
|---|---|
| 2026-07-31 | 613 |
| 2026-08-02 | 613 |
| 2026-08-06 | 613 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
<div align="center">
<h1>Q-Align: Teaching LMMs for Visual Scoring via Discrete Text-Defined Levels</h1>
See you in ICML 2024, Vienna (Wed 24 Jul 1:30 a.m. CET — 3:00 am CET)
<div style="width: 50%; text-align: center; margin:auto;">
<img style="width: 50%" src="fig/onescorer.png">
</div>
<a href="https://huggingface.co/spaces/teowu/OneScorer">One Unified Model for Visual scoring.</a>
<div>
<a href="https://teowu.github.io/" target="_blank">Haoning Wu</a><sup>1</sup><sup>*</sup><sup>+</sup>,
<a href="https://zzc-1998.github.io/" target="_blank">Zicheng Zhang</a><sup>2</sup><sup>*</sup>,
<a href="https://sites.google.com/view/r-panda" target="_blank">Weixia Zhang</a><sup>2</sup>,
<a href="https://chaofengc.github.io" target="_blank">Chaofeng Chen</a><sup>1</sup>,
<a href="https://liaoliang92.github.io" target="_blank">Liang Liao</a><sup>1</sup>,
<a href="https://github.com/lcysyzxdxc" target="_blank">Chunyi Li</a><sup>2</sup>,
</div>
<div>
<a href="https://github.com/YixuanGao98" target="_blank">Yixuan Gao</a><sup>2</sup>,
<a href="https://github.com/AnnanWangDaniel" target="_blank">Annan Wang</a><sup>1</sup>,
<a href="https://github.com/ZhangErliCarl/" target="_blank">Erli Zhang</a><sup>1</sup>,
<a href="https://wenxiusun.com" target="_blank">Wenxiu Sun</a><sup>3</sup>,
<a href="https://scholar.google.com/citations?user=uT9CtPYAAAAJ&hl=en" target="_blank">Qiong Yan</a><sup>3</sup>,
<a href="https://sites.google.com/site/minxiongkuo/" target="_blank">Xiongkuo Min</a><sup>2</sup>,
<a href="https://ee.sjtu.edu.cn/en/FacultyDetail.aspx?id=24&infoid=153&flag=153" target="_blank">Guangtao Zhai</a><sup>2</sup><sup>#</sup>,
<a href="https://personal.ntu.edu.sg/wslin/Home.html" target="_blank">Weisi Lin</a><sup>1</sup><sup>#</sup>
</div>
<div>
<sup>1</sup>Nanyang Technological University, <sup>2</sup>Shanghai Jiao Tong University, <sup>3</sup>Sensetime Research
</div>
<div>
<sup>*</sup>Equal contribution. <sup>+</sup>Project Lead. <sup>#</sup>Corresponding author(s).
</div>
<div>
<a href="https://HuggingFace.co/q-future/one-align"><strong>OneAlign (HF Model)</strong></a> | <a href="https://github.com/Q-Future/Q-Align/tree/main/model_zoo"><strong>Model Zoo</strong></a> | <a href="https://q-future.github.io/Q-Align/fig/Q_Align_v0_1_preview.pdf"><strong>Technical Report</strong></a> | <a href="https://q-align.github.io/"><strong>Homepage</strong></a> | <a href="https://huggingface.co/spaces/teowu/OneScorer"><strong>OneScorer (HF Space)</strong></a>
</div>
<div>
**The retrained version on Qwen3.5-VL can be found at [Q-ReAlign](https://github.com/Q-Future/Q-ReAlign)**
<h2>Results</h2>
<div style="width: 75%; text-align: center; margin:auto;">
<img style="width: 75%" src="fig/radar.png">
</div>
<h2>Syllabus</h2>
<div style="width: 100%; text-align: center; margin:auto;">
<img style="width: 100%" src="fig/q-align-syllabus.png">
</div>
<h2>Structure</h2>
<div style="width: 100%; text-align: center; margin:auto;">
<img style="width: 75%" src="fig/structure.png">
</div>
</div>
## [New Update]
1. We propose the first models (Q-SiT, Q-SiT-mini) that can deal with both image quality scoring and interpreting tasks. See more details [here](https://github.com/Q-Future/Q-SiT).
2. **We release the new version of Q-Align, [Q-ReAlign](https://github.com/Q-Future/Q-ReAlign), which is trained on modern VLMs and gains the same performance with 50% fewer parameters.**
## [Important Note!] v1.1 <-> transformers==4.36.1+
This LLaVA-style repository has been built on `transformers==4.31.0`, which is incompatible with many new models available on Hugging Face. This requires to build a separate environment for the MLLM/LMM repository, which is somewhat troublesome for this **visual scoring** model, as we expect the **Q-Align/OneAlign** to effectively boost other disciExcerpt of 11,042 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:b3d137be00bab011, desc:scoring
