Ph.D. candidate @NUSingapore | Efficient ML | ACL Outstanding Paper Award
Singapore
Joined December 2021
- Tweets19
- Following38
- Followers125
- Likes77
Pinned Tweet
Training-free Video Enhancement: Achieved 🎉
Nice work with @oahzxl @shaowenqi126301 @VictorKaiWang1 @VitaGroupUT @YangYou1991 et al.
Non-trivial enhancement, training-free, and plug-and-play 🥳
Blog: oahzxl.github.io/Enhance_A_V… (🧵1/6)
Amazing work!
Is model scaling the only source of agent improvement?
We @henryqin1997 @YaxinLu1997 @VITAGroupUT @VictorKaiWang1 are glad to share our work: We reach 95.3% Raw Accuracy, or a $15 Frontier Run (88.8% with DeepSeek-v4-Flash, matching GPT-5.6 Sol Max), on Terminal-Bench 2.1 via Harness Scaling.
Human effort -> Non scalable ❌
Agent direction? Local optimization ❌
Human direction + agent action = better workflow control✅
StateM:
1. Open-source model + StateM -> Close-source frontier, everyone can try!
2. Training free, significantly improve frontier agents.
3. Zero-transfer-cost on same family of agents. [1/8]🧵
📄arxiv.org/abs/2608.15089
🌐 henryqin1997.github.io/state…
Amazing work on parameter generation!!
One static model does not fit all😭
We just dropped our latest work: Functional Neural Memory. Instead of static models, we generate custom "parameters" for every single input.
✅Prompt your model anytime
✅Instant personalization
✅Better instruction following
✅Flexible & dynamic memory (w/o memory bank✌️)
(🧵1/6)
Yang Luo retweeted
Customizing Your LLMs in seconds using prompts🥳!
Excited to share our latest work with @HPCAILab, @VITAGroupUT, @k_schuerholt, @YangYou1991, @mmbronstein, @damianborth : Drag-and-Drop LLMs(DnD). 2 features: tuning-free, comparable or even better than full-shot tuning.(🧵1/8)
Yang Luo retweeted
Representation Alignment (REPA) is NOT ALWAYS helpful for diffusion training!🤷
Sharing latest work w/ @HPCAILab and @VITAGroupUT: "REPA Works Until It Doesn't: Early-Stopped, Holistic Alignment Supercharges Diffusion Training".
Acceleration up to 28x w/o performance drop.(🧵1/7)
Glad to see CAME's efficiency in the training of SANA 1.5!
Yang Luo retweeted
🚀Towards efficient Diffusion Transformers!
😆We are happy to introduce RAS, the first diffusion sampling strategy that allows for regional variability in sampling ratios, achieving up to 2x+ speedup!
🔌Training-free, plug and play!
💪Nice work with @MSFTResearch @YangYou1991 @Yif_Yang et al.
📜Paper: huggingface.co/papers/2502.1…
📖Blog: aka.ms/ras-dit
⌨️Code: github.com/microsoft/RAS (1/5)
Yang Luo retweeted
Generating ~200 million parameters in just minutes! 🥳
Excited to share our work with @MTDovent , @heisejiasuo96 , and @YangYou1991: 'Recurrent Diffusion for Large-Scale Parameter Generation' (RPG for short).
Example: Obtain customized models using prompts (see below). (🧵1/8)
For more details, check out our blog: oahzxl.github.io/Enhance_A_V…
Our code is open-source on GitHub: github.com/NUS-HPC-AI-Lab/En…
More results and the technique report will be coming soon! (🧵6/6)
We are pleased to announce that PIXART-Σ is trained by our CAME optimizer(arxiv.org/abs/2307.02047).
Glad to see that our work has a real-world impact on the training of DiT models!
Code: github.com/yangluo7/CAME
Yang Luo retweeted
Explore the blog post for a concise and insightful overview of the CAME optimizer! Congrats to the first author @YangYoungLL! 🔥
Blog: zhengzangw.github.io/blogs/c…
ArXiv: arxiv.org/abs/2307.02047
Code: github.com/huawei-noah/Pretr… (a plug-and-play optimizer repo will be released soon)
🎉 We're thrilled and honored to have received the ACL Outstanding Paper Award! 🏆 Stay tuned, as we'll be publishing a repository for the CAME optimizer very soon. #ACL2023