Yang Luo

Ph.D. candidate @NUSingapore | Efficient ML | ACL Outstanding Paper Award

[Yang Luo @YangL_7]

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: https://t.co/8Cz78M0L7v (🧵1/6)

[Yang Luo @YangL_7]

Amazing work on parameter generation!!

[Tencent Hy @TencentHunyuan]

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 @YangL_7]

Customize your LLM in seconds!

[AK @_akhaliq]

Drag-and-Drop LLMs

Zero-Shot Prompt-to-Weights

[Victor.Kai Wang @VictorKaiWang1]

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)

[Ziqiao Wang @ZiqiaoWang63428]

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)

[Yang Luo @YangL_7]

Glad to see CAME's efficiency in the training of SANA 1.5!

[Enze Xie @xieenze_jr]

🔥 SANA 1.5: A linear Diffusion Transformer pushes SOTA in text-to-image generation!

Key innovations:
• Depth-growth training: 1.6B → 4.8B params
• Memory-efficient 8-bit optimizer
• Flexible model pruning
• Inference scaling for better quality

Achieves 0.80 on GenEval! ����

[Ziming Liu @lzm_mlsys]

🚀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: https://t.co/IHDk6ihhMz
📖Blog: https://t.co/KZ0eoje3YN
⌨️Code: https://t.co/LtzcJyHJiS (1/5)

[Victor.Kai Wang @VictorKaiWang1]

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)

[Yang Luo @YangL_7]

@nirzabari @oahzxl @shaowenqi126301 @VictorKaiWang1 @VITAGroupUT @YangYou1991 Thank you! Glad to hear the improvements worked well with LTXV too. Your observation about the enhanced coherence in movement and detail preservation is super helpful feedback.

[Yang Luo @YangL_7]

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: https://t.co/8Cz78M0L7v (🧵1/6)

[Yang Luo @YangL_7]

@ai_kimera @oahzxl @shaowenqi126301 @VictorKaiWang1 @VITAGroupUT @YangYou1991 Thanks! Really appreciate you noticing the progress, and we're excited to keep pushing forward.

[Yang Luo @YangL_7]

For more details, check out our blog: https://t.co/j1dI0ahdjJ
Our code is open-source on GitHub: https://t.co/gRa5P7kjfA
More results and the technique report will be coming soon! (🧵6/6)

[Yang Luo @YangL_7]

Our approach: we compute the average of non-diagonal elements of temporal attention maps as cross-frame intensity (CFI). An enhanced temperature parameter multiplies the CFI to enhance the temporal attention output.

[Yang Luo @YangL_7]

@oahzxl Lei God

[Yang Luo @YangL_7]

We are pleased to announce that PIXART-Σ is trained by our CAME optimizer(https://t.co/Rv06olptwD).

Glad to see that our work has a real-world impact on the training of DiT models!

Code: https://t.co/ROCS7y2wF2

[Zangwei Zheng @ZangweiZheng]

Explore the blog post for a concise and insightful overview of the CAME optimizer! Congrats to the first author @YangYoungLL! 🔥

Blog: https://t.co/jPStEQg1ja
ArXiv: https://t.co/lRuvsQUwnU
Code: https://t.co/6rEiKttuhe (a plug-and-play optimizer repo will be released soon)