Shuo Yang

@Andy_ShuoYang

3rd year phd at Berkeley; Efficient ML System;

Berkeley

andy-yang-1.github.io

Replies Reposts Media

Pinned
Shuo Yang
@Andy_ShuoYang
Aug 21
Your gaming PC can now serve frontier models at interactive speed using official checkpoints without extreme quantization! Qwen3.6 35B → 8GB RTX 4060 laptop @ 39 tok/s DeepSeek-V4-Flash 284B → RTX 5090 desktop @ 22-25 tok/s GLM-5.2 753B → RTX PRO 6000 workstation @ 15 tok/s
00:00
260
Shuo Yang
@Andy_ShuoYang
Sep 3
Congrats!
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
Shuo Yang
@Andy_ShuoYang
Sep 3
Qwen3.8-Flash-Next now runs on a single RTX 5090 at 68.3 tok/s — no extreme quantization, no speculative decoding. · Production-level checkpoint: GB300-validated NVFP4 checkpoint from RadixArk · 63GB host RAM — less than what 1-bit quants of this model need · The 51GB n-gram
00:00
37
Shuo Yang
@Andy_ShuoYang
Sep 3
Linear attention is playing a more important role in videogen. Nice work @HaochengXiUCB!
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
Shuo Yang
@Andy_ShuoYang
Aug 4
flashlib + CAKE: geomean 1.9x faster exact KNN on Blackwell 🚀 NVIDIA's CAKE team contributed generated CUDA kernels (tcgen05, per-shape dispatch) — 1.9x geomean / 5.4x peak across 198 shapes vs the previous flashlib release, zero recall loss. Live on flashlib nightly
Refresh KNN-search standalone export from the Cake MR 415 head (E2E 1.899x) by yyihuang · Pull...
From github.com
4