Argilla

You can now edit datasets directly on the Hugging Face Hub. The future of datasets for AI is here. 🔥 No more downloading a 500MB CSV just to fix 3 mislabeled rows. The workflow: 1. Spot an error in your dataset 2. Click edit in Data Studio 3. Fix the cells 4. Commit with a message 5. Repeat! Every change is versioned like code. Full traceability. The interesting part: collaborative curation. Your team can make commits to the same dataset, review each other's changes, and improve data quality together. I just published a blog post with a practical example: https://lnkd.in/d3JVvvh9 This will change how we maintain datasets for AI. Please leave a comment with your feedback and ideas to help us shape the future.

Looking for the best open model for your use case? Want to evaluate different inference providers? Search no more I am excited to announce the integration of Inference Providers into the fastest-growing eval framework, used by Anthropic, Groq, and many others. Run evals across thousands of open models and inference providers with Hugging Face Inference Providers and InspectAI. Get started https://lnkd.in/diCQqK_M

🚀 Launching a new tutorial: Extract text and knowledge from your photos with Open Vision Language Models With the recent surge of OCR open models, there has been a new wave of content and tutorials for document-oriented use cases. In this blog post, I focus on using general VLMs in less structured images, like handwritten recipes, where you want to perform several transformations, ask questions, and combine them with LLMs. https://lnkd.in/d9JDSEnz

What do you do with the photos you take to remember something? I usually snap quick photos at exhibitions, museums, or conferences to keep track of what interests me. It could be a slide, a poster, a piece of art, or even a short note I want to remember. With AI Sheets, I can organize these fragments and turn them into practical knowledge instead of letting them pile up. With just a few clicks and short prompts, I can extract text from the images (even the hard-to-read ones) and add context about the artwork. The result is a structured spreadsheet: a personal knowledge base built without endless prompt iterations. Another advantage is that AI Sheets can run different open-source models within the same workflow. Some models handle complex tasks like text extraction or image interpretation better, while others are more efficient for simpler tasks such as translation or classification. Curious to see how your photos might be useful to you :) We wrote a post about it in case you want to explore it further and try it out. https://lnkd.in/dP53dsZ4

🔥 Excited to announce Hugging Face AI Sheets 2.0: Extract text, detect objects, edit, and classify images with vision and language models AI Sheets is an open-source tool for supercharging datasets with open AI models, no code required. Now with vision support: extract data from images (receipts, documents), generate visuals from text, and edit images—all in a spreadsheet. Powered by thousands of open models via Inference Providers. Read all about it: https://lnkd.in/dDK7kzmR

🔥 Let's go! Combining LLMs, Vision Language Models, and text-to-image, with the best open models on Hugging Face. Starting today, you can use VLMs directly on Hugging Face AISheets. Try it: https://lnkd.in/gdKeV-zW Here's an example: One of the most exciting use cases of VLMs is assisting people who are blind or have low vision. I asked myself the following: Can I use VLMs to describe artworks helpfully? If so, how can I test whether the descriptions are accurate or useful without reading them all? So I built the following workflow without writing a line of code: 1. Import the wikiart dataset from HF 2. Write a prompt to describe the image using Qwen-2.5-VL 3. Use the description to generate images using Qwen-Image The results are pretty surprising and fun! Stay tuned in the coming days, I'll share the dataset and prompts to reproduce it.

🔥 500,000 public datasets on Hugging Face! Perfect time to ask the community 👇 What are we missing to help you build and share more and higher-quality datasets?