Chenfeng_X

@Chenfeng_X

Assistant Professor @UTCompSci, PhD @UCBerkeley. Working on something really works

Berkeley, California

chenfengx.com

Replies Reposts Media

Pinned
Chenfeng_X
@Chenfeng_X
May 19
Excited that our paper StreamdiffusionV2 received the Best Research Paper Award at #MLSys26! 🚀Video generation is quickly moving from demos to production-facing workloads. It is no longer a turn-based pipeline but should be a streaming pipeline to interact with users. 📖Our
00:00
5
Chenfeng_X
@Chenfeng_X
Sep 4
Congrats @lmxyy1999 and the team!
Nunchux AI
@NunchuxAI
Sep 3
Nunchux is live at nunchux.ai! We’re building the frontier of multimodal generative AI inference: fast, affordable, high-quality serving for image, video, and world models. Our Modelverse brings 30+ image, video, and avatar models behind one API. Use
Chenfeng_X
@Chenfeng_X
Sep 3
Great work from @HaochengXiUCB, @HavenFeng and the team. Near-lossless quality drop, livestream-speed serving. Congrats! 👏
Haocheng Xi
@HaochengXiUCB
Sep 2
Open-source video generation is now faster than playback without compromising quality. Introducing Video Delta Net (VDN): hybrid attention for live text-to-video with near-lossless quality. VDN accelerates Minimax-H3 by 75 - 90 x, generating 14 seconds of 768p video in 11
00:00
1
Chenfeng_X
@Chenfeng_X
Jun 12
Our work carefully discussed how to loop effectively for dLLM 🙂 check @yuezhouhu ‘s post below 👇
Yuezhou Hu
@yuezhouhu
Jun 12
Very excited to see that the core idea of DiffusionGemma directly stems from our work, Residual Context Diffusion (arXiv:2601.22954)! Code- and architecture-level comparisons are attached. 🏆 RCD is accepted to ICML 2026! See you in Seoul! #DiffusionLLM #LLM #Reasoning #GenAI
Chenfeng_X
@Chenfeng_X
Jun 3
LEANN (github.com/StarTrail-org/…)+ FlashLib (flashml-org.github.io) = 26x faster LEANN. Looking forward to more X + FlashLib!
Shuo Yang
@Andy_ShuoYang
Jun 3
FlashLib update: we now support ANN search with IVF-Flat — up to 6.5× faster than cuVS on real-world vector workloads (SIFT-1M) while matching recall. LEANN now supports FlashLib as a backend: 26× faster build, 29× faster single-query, and 298× faster batch search. Huge thanks
00:00
1