Hugging Face

Hello! 📢 Newcomer in the NuExtract family 📢 We just released NuExtract 1.5 smol, based on Hugging Face SmolLM v2 1.7B. A small model, but don't get fooled by its size, the thing can extract! https://lnkd.in/e6-P6jTu In more details: We have been repeatedly asked for small versions of NuExtract. We previously released NuExtract 1.5 tiny, based on Qwen 2.5 0.5B, but unfortunately, we could not get it to: - be multilingual - have continuation abilities (extracting while being given previously extracted information) Recently Hugging Face released SmolLM 2 1.7B, which has excellent benchmark performance, so we decided to give it a go. And it turns out to work pretty well :) We managed to give it all the abilities that NuExtract 1.5 has, and it reaches performance similar to the original NuExtract while being less than half its size! Hope you will make good use of this model, and as always don't hesitate to give us feedback here and in our Discord server: https://lnkd.in/ebAD8NJ6. NuExtract 2.0 is cooking and we want to make it as good as we can! -- model creator: Liam Cripwell, and thanks Leandro von Werra, Daniel van Strien, and Loubna Ben Allal for early access to SmolLM 2 :)

Introducing gradio's new component -- gr.BrowserState() 🤩 Gradio apps now support local state, where data persists in the browser's local-Storage even after the page is refreshed or closed. 🚀 Store user preferences, 🚀 Settings, 🚀 API keys, or 🚀 Anything you want to persist across sessions! The attached demo shows how users can apply themes to their apps, and their choice persists even after refreshes or restarts of the app! 🤯 🔥 Access this Theme Customization Demo on Hugging Face Spaces here: https://lnkd.in/gdBu7XpH You can also use gr.BrowserState() to save your passwords or API keys across the sessions🤯. For more information on how to use `gr.BrowserState()`, please refer to our documentation: https://lnkd.in/gUzGGiGb Upgrade to our latest Gradio release 5.6.0: pip install gradio --upgrade

Mochi-1 is now supported in `diffusers` natively. Please refer to its original model card for more details. Reminder: It's an Apache 2.0 model! Thanks to the Genmo team for making their amazing work open!

Mochi-1 is now supported in `diffusers` natively. Please refer to its original model card for more details. Reminder: It's an Apache 2.0 model! Thanks to the Genmo team for making their amazing work open!

`bitsandbytes` makes it really easy to quantize your models. `transformers` and `diffusers` have the backend baked in. This means you would need less that 16GB of GPU VRAM to run Flux Dev. Note: The diagram states MB for memory, which should be GB (sorry for the mistake)

Play with the viral "Text Behind Image" tool, recreated with Gradio and Birefnet-lite! 🔥 The first version is now live. You can quickly generate as many images as you want 🚀 Play with the app on Hugging Face 🤗 Spaces! Link 👇 https://lnkd.in/gEhx_97H

AutoTrain repo has 4k stars now 🔥 https://lnkd.in/dUmxgcUX

The new Qwen-2.5-Coder is 🔥🔥🔥 and this is how to test it and improve it for your use case Say you want a highly capable Rust coding assistant. You need to evaluate Qwen for this, improve the prompt, and potentially fine-tune it. So we’ve built this Colab to get you started, end-to-end: 1. Build a synthetic code dataset for Rust with distilabel. 2. Use this dataset to do human evaluation with Argilla. All on the Hugging Face Hub: open source tools and models FTW! Colab: https://lnkd.in/dxVFHvKD Next step: use correct responses as few-shot examples to make responses more accurate.

Qwen 2.5-Coder is now available on Hugging Chat 🤗 - GPT-4 coding capabilities - Open source / Apache 2.0 - 40+ programming languages supported - Available in 6 sizes (0.5B-32B)

Wow! Qwen 2.5 Coder has been released with code artifacts! 🤯 Qwen has open-sourced the powerful and diverse Qwen2.5-Coder Series (0.5B/1.5B/3B/7B/14B/32B). Apache 2.0 license! The code artifacts app is built with Gradio 5 and makes use of our custom components feature too 🤩 Qwen2.5-Coder apps are live on Hugging Face Spaces! 💪 Qwen2.5-Coder Instruct demos: https://lnkd.in/ga9GeFNy 👍 Qwen2.5-Coder 7B Instruct: https://lnkd.in/gsD3VT7S 🔥🔥🔥 The Code-Artifacts app that uses Qwen2.5-Coder 32b-instruct: https://lnkd.in/gcg9gpEd