Hugging Face

Inference-time scaling meets Flux.1-Dev (and others) 🔥 Presenting a simple re-implementation of "Inference-time scaling diffusion models beyond denoising steps" by Willis (Nanye) Ma et al. I did the simplest random search strategy, but results can potentially be improved with better-guided search methods. Supports Gemini 2 Flash & Qwen2.5 as verifiers for "LLMGrading" 🤗 The steps are simple: For each round: 1> Starting by sampling 2 starting noises with different seeds. 2> Score the generations w.r.t a metric. 3> Obtain the best generation from the current round. If you have more compute budget, go to the next search round. Scale the noise pool (2 ** search_round) and repeat 1 - 3. This constitutes the random search method as done in the paper by Google DeepMind. Code, more results, and a bunch of other stuff are in the repository. Check it out here: https://lnkd.in/gPRDz7KB 🤗 Thanks to Willis for all the help in getting this shipped!

NEW COURSE! We’re cooking hard on Hugging Face courses, and it’s not just agents. The NLP course is getting the same treatment with a new chapter on Supervised Fine-Tuning! 👉 Follow to get more updates https://lnkd.in/eXZPaJZT The new SFT chapter will guide you through these topics: 1️⃣ Chat Templates: Master the art of structuring AI conversations for consistent and helpful responses. 2️⃣ Supervised Fine-Tuning (SFT): Learn the core techniques to adapt pre-trained models to your specific outputs. 3️⃣ Low Rank Adaptation (LoRA): Discover efficient fine-tuning methods that save memory and resources. 4️⃣ Evaluation: Measure your model's performance and ensure top-notch results. This is the first update in a series, so follow along if you’re upskilling in AI.

🎯 Breaking: Pure Vision-Based GUI Interaction is Here Microsoft's OmniParser V2 - missing link between computer vision and UI automation. What it does: - Converts GUI screenshots into structured elements - Enables natural interaction patterns for GPT-4V - Works with any interface, no rules required - Pure vision-based understanding This is the solid foundation that vision-based AI agents have been waiting for. Kudos to Yadong (Adam) Lu and rest of the team at Microsoft. Want to see the future of AI-UI interaction? Explore with the app on Hugging Face Spaces (Useful links are in the top comment)

😉 What if your phone could talk directly to AI? Not through an app or a website. But.. direct communication. We're about to change how developers build real-time AI communication. Python devs, you might want to watch this space very closely over the next few days for our upcoming launch 👀🔥🔥

I'm super excited about the close partnership with Hugging Face 🤗 🔥 ! We share the same goal of bringing the best open models and AI developer tools to the community! Today, our fast Fireworks AI inference engine is available on the Hugging Face Hub. We enabled it quickly per community requests. You can access Deepseek R1/v3, llama3.3 and Mistral and many other models served by Fireworks. We are thrilled to continue to provide the best AI tools from Fireworks to the broad Hugging Face community. Thank you, Julien Chaumond and the Hugging Face team, for the Valentine's Day gift! 👉 Detailed post: https://lnkd.in/eWhvtyKq

Fireworks AI is now a supported Inference Provider on Hugging Face 🎆 This means Fireworks delivers blazing-fast serverless inference: - directly on model pages, - as well as throughout the HF ecosystem of libraries and tools, including the tools from the community. This makes it easier than ever to run inference on your favorite models. On a personal note, it's been awesome working with Lin Qiao and her team on this ❤️. Happy valentine's day!

Deepseek R1 just became the most liked model ever on Hugging Face just a few weeks after release - with thousands of variants downloaded over 10 million times now!

Yes!

𝗚𝗿𝗲𝗮𝘁 𝗳𝗲𝗮𝘁𝘂𝗿𝗲 𝗮𝗹𝗲𝗿𝘁: you can now share agents to the Hub! 🥳🥳 And any agent pushed to Hub get a cool Space interface to directly chat with it. This was a real technical challenge: for instance, serializing tools to export them meant that you needed to get all the source code for a tool, verify that it was standalone (not relying on external variables), and gathering all the packages required to make it run. Go try it out! 👉 https://lnkd.in/euFmXyGS

🚀BREAKING: AI Development Just Got 10x Faster! Don't spend hours digging into docs to build AI app. Introducing Playground Search for Gradio - your new AI development superpower: ✨ Instant documentation discovery ✨ Smart code generation ✨ Real-time demo suggestions Want to see the magic? Just type "How do I use the image component?" and watch what happens. 🤩 This isn't just another update. This is the democratization of AI development.