DeepSeek-Coder-V2 is out! 🔥236B Mixture of Experts with 21B active experts 🧠6 trillion tokens ⌨️338 programming languages 📏128k context length 🤏Tiny version also released (16BA2.4B) 🕴️Code completion/infilling/chat Models: huggingface.co/collections/de…

This is no Woodstock AI but will be fun nonetheless haha. I’ll be hosting a live workshop with team members next week about the Enterprise Hugging Face hub. 1,000 spots available first-come first serve with some surprises during the stream!

Don't tell the hub team but I found a hack in enterprise hub subscriptions that I'll share during the live workshop Wed 😅😅😅🤑🤑🤑

Massachusetts Institute of Technology (MIT) Roy Shilkrot I made a collection of all the papers I will talk about here, each paper page is linked to their models and demos too in case you want to try them ☺️ huggingface.co/collections/me…

really cool, great job 🤗

🚀 Excited to introduce TritonAcademy – a new GitHub repository for the ML/AI community! After receiving great feedback on my Cutlass work, I’ve had many requests for Triton resources. So, I’m building an educational repo to demystify famous Triton kernels from Unsloth, Liger,

Some people today are discouraging others from learning programming on the grounds AI will automate it. This advice will be seen as some of the worst career advice ever given. I disagree with the Turing Award and Nobel prize winner who wrote, “It is far more likely that the

The Hidden Trillions: The True Value of Open Source Software Harvard’s latest 42 pages study reveals OSS is worth trillions but remains massively underfunded. 🔹 96% of codebases rely on OSS 🔹 Recreating OSS from scratch? $8.8T+ 🔹 Just 5% of contributors create 96% of the

We've announced cuTile, a tile programming model for CUDA! It's an array-based paradigm where the compiler automates mem movement, pipelining & tensor core utilization, making GPU programming easier & more portable. I'm proud of my stellar team for all their hard work on this!

Without any resource use on your side 🤗 👾 quantize and push to the hub any model through this cool new bitsandbytes community space. Thanks Mohamed and Marc Sun for creating this useful new community space! 🔥🙏

‼️Sentence Transformers v4.0 is out! You can now train and finetune reranker (aka cross-encoder) models with multi-GPU training, bf16 support, loss logging, callbacks & much more. I also prove that finetuning on your domain helps much more than you might think. Details in 🧵

Check out the latest release of 🤗 Accelerate! This update is packed with exciting features, including the much-anticipated FSDPv2 integration. Discover why FSDPv2 is a game-changer in the thread below. 🧵

If you’re excited about optimizing code that runs equally well on a single or thousands of GPUs and if you have the ability to submit a single substantial PR to a major OSS library, we want you on the PyTorch team - especially if you’re early in your career.

We've just revamped the @Huggingface Quantization docs! 🥳 Understand concepts better & choose the right technique for your needs with these key updates: - Explanations of quantization fundamentals (schemes, int4, FP8). huggingface.co/docs/transform… New Selection Guide: Choose the

A quick update on the future of the `transformers` library! In order to provide a source of truth for all models, we are working with the rest of the ecosystem to make the modeling code the standard. A joint effort with vLLM, LlamaCPP, SGLang, Mlx, Qwen, Glm, Unsloth, Axoloth,

The Transformers library is undergoing it's largest pivot to date 🙌 It now cements its role as the central model definition, irrespective of the backend and runner. One ground truth to bring more reliability across the ecosystem. Why is this important?

New Blog Post! 🚀 Explore how quantization backends in Diffusers make large diffusion models like Flux run with less VRAM without sacrificing (much) quality! Run Flux with bitsandbytes-4bit using under 18 GB VRAM and in just 15 seconds! blogpost: huggingface.co/blog/diffusers… Can you

Bitsandbytes latest works with `torch.compile(fullgraph=True)` and you should put it to good use 🔥 For example, when applied to Flux, it beefs up the performance quite a bit. Code: gist.github.com/sayakpaul/0db9… Enjoy 🔥

Today is a big day, we're introducing the first version of the HF MCP server 🔥 🧵

My sleep scores during recent travel were in the 90s. Now back in SF I am consistently back down to 70s, 80s. I am increasingly convinced that this is due to traffic noise from a nearby road/intersection where I live - every ~10min, a car, truck, bus, or motorcycle with a very