Log inSign up
Log inSign up
Matthew Douglas
340 posts
Matthew Douglas profile banner
@mattkdouglas

Matthew Douglas

@mattkdouglas
ML Engineer 🤗 @huggingface
Indianapolis, IN
Joined May 2011
514 Following
338 Followers
RepliesRepliesRepostsRepostsMediaMedia

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.
  • @mattkdouglas
    Matthew Douglas
    @mattkdouglas
    Sep 3
    Very exciting news! Was awesome to have @JensenHuang on our company call this morning!
    @ClementDelangue
    clem 🤗
    @ClementDelangue
    Sep 3
    Super happy to share our intention to join forces with NVIDIA in a $12,930,300,000 acquisition 💛💚 10 years after starting Hugging Face, open-source AI is at an inflection point. Thanks to the community, we’ve shown that it can be a complement, and even an alternative, to
  • @mattkdouglas
    Matthew Douglas
    @mattkdouglas
    Jul 28
    bitsandbytes v0.50.0 is out! 4bit LLM inference is up to 4x faster on NVIDIA GPUs! And the speedups aren't only on NVIDIA this time. AMD ROCm also gets a new fast 4bit path, and is now stable! Plus, builds are included in Windows wheels and now support more GPUs.
    2
  • @mattkdouglas
    Matthew Douglas
    @mattkdouglas
    Mar 14, 2025
    Exciting times for open-source!
    @allen_ai
    Ai2
    @allen_ai
    Mar 13, 2025
    Announcing OLMo 2 32B: the first fully open model to beat GPT 3.5 & GPT-4o mini on a suite of popular, multi-skill benchmarks. Comparable to best open-weight models, but a fraction of training compute. When you have a good recipe, ✨ magical things happen when you scale it up!
  • @mattkdouglas
    Matthew Douglas
    @mattkdouglas
    Dec 6, 2024
    If you're upgrading @huggingface 🤗 Accelerate to 1.2.0 and use bitsandbytes, make sure to move your model and inputs to your chosen device. Otherwise, you may see a RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!.
    Code example demonstrating a RuntimeError when device placement is not specified.
    3
  • @mattkdouglas
    Matthew Douglas
    @mattkdouglas
    Dec 5, 2024
    Announcing the bitsandbytes 0.45.0 release! 🆕 H100 support for LLM.int8() 🚀 Much faster LLM.int8() inference 🚀 Faster 4bit inference 📦 Reduces Linux wheel size from (122.4MB to 69.1MB) Release notes: github.com/bitsandbytes-f…
    @Tim_Dettmers
    Tim Dettmers
    @Tim_Dettmers
    Dec 5, 2024
    The new bitsandbytes is here: ~15% faster 4-bit ~70% faster 8-bit inference 8-bit support for H100s github.com/bitsandbytes-f… Great engineering from @mattkdouglas. bitsandbytes now receives about 100,000 installations daily. A little history on 8-bit implementations in bnb 🧵
    1