1. X
  2. Xiangtai Li
Log inSign up
Xiangtai Li
310 posts
Xiangtai Li profile banner
user avatar

Xiangtai Li

@xtl994
Research Scientist@ByteDance Inc. SG, Post-Doc in MMLab@NTU, PhD@PKU. Opinions are my own.
Singapore
lxtgh.github.io
Joined December 2022
762
Following
970
Followers
RepliesRepliesMediaMedia
  • user avatar
    Xiangtai Li
    @xtl994
    Aug 5
    学习下
    user avatar
    Kimbo
    SemiAnalysis
    @kimbochen
    Aug 3
    Replying to @kimbochen
    Songlin Yang's video explanations youtube.com/watch?v=d0HJvG… youtube.com/watch?v=RTJKXK… Songlin Yang's blog post Design intuition sustcsonglin.github.io/blog/2024/delt… Kernel algorithm (I skipped to "A Chunkwise Algorithm for DeltaNet" section) sustcsonglin.github.io/blog/2024/delt… FlashKDA
  • user avatar
    Xiangtai Li
    @xtl994
    Jun 27
    fluent chinese 哈哈哈哈
    user avatar
    Zizheng Pan
    @zizhpan
    Jun 26
    We’re expanding our team! Come and join us! You will be working in Beijing or Hangzhou. Fluent Chinese is required. 各个岗位热招,均可实习,欢迎投递🙌
    Image
    Image
  • user avatar
    Xiangtai Li
    @xtl994
    May 18
    very cool guide👍
    user avatar
    Jia-Bin Huang
    @jbhuang0604
    May 17
    Modern Transformer - Complete Guide Interested in learning the recent advances in transformers? After 13 videos, I've finally completed this series! 🥳🥳🥳 Check out the course here: youtube.com/playlist?list=…
    Image
  • user avatar
    Xiangtai Li
    @xtl994
    Apr 8
    🐮
    user avatar
    Pierre-Antoine Bannier
    @el_PA_B
    Apr 8
    sam3.cpp - Meta's SAM 3 in pure C++ with @ggerganov's ggml - Supports SAM 3.1, 3, 2.1, 2 and EdgeTAM - FP16, 4-bit quant (EdgeTAM in 15 MB) - Apple Metal GPU, CUDA, CPU - Text-prompted: "peach" → every peach - Single-file C++14 Performance-wise: - 100ms object detection,
    Image
    00:00
    Image
    00:00
  • user avatar
    Xiangtai Li
    @xtl994
    Mar 6
    Looks like a super hyper-network for foundation models, interesting.
    user avatar
    Tencent Hy
    @TencentHunyuan
    Mar 5
    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 &
    Image
    00:00

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Advertisement
Advertisement