nigel joseph @tpdevelopment

This is HUGE, Hugging Face just shipped Inference Providers on the Hub Partnering with @togethercompute, @fal, @replicate and @SambaNovaAI! Starting today you can access thousands of Models like DeepSeek R1, Llama, Flux, Whisper, and more Directly from Hugging Face!

We have authored a post to go over the state of video generation in the Diffusers ecosystem 🧨 We cover the models supported, the knobs of optims our users can fire, fine-tuning, and more 🔥 5-6GBs for HunyuanVideo, sky is the limit 🌌 🤗

Introducing Bespoke-Stratos-32B, our reasoning model distilled from DeepSeek-R1 using Berkeley NovaSky’s Sky-T1 recipe. The model outperforms Sky-T1 and o1-preview in reasoning (Math and Code) benchmarks and almost reaches the performance of DeepSeek-R1-Distill-Qwen-32B while being trained on 47x fewer examples! Crucially, we open-source the dataset (DeepSeek open-sourced the model, not the data). Let's work together on this exciting direction of reasoning distillation! 🧵More info and link to the blog below!

Introducing "Flux-Edit", an experimental finetune of Flux.1-Dev (@bfl_ml), following the "Flux Control" framework 🔥 Works nicely with a wide variety of editing tasks beyond style transfer. Works with turbo LoRAs, reducing steps from 50 to 8. Ckpt + code ⬇️

New model alert! 🚨 ⋆✴︎˚｡FLEX.1 Alpha ˚｡✴︎⋆ is an 8B parameter model pruned and further trained by @ostrisai from 12B FLUX.1 [schnell]: 🖼️ High quality, competitive with FLUX[dev] 🎨 Good at styles 🤏 Smol 📜 openly licensed (Apache 2.0) ⚗️ de-destiled, CFG optional

DeepSeek-R1 GGUF's are now on @HuggingFace! Includes all Llama & Qwen distilled models + 2 to 8-bit quantized versions. How to run R1: unsloth.ai/blog/deepseek-… DeepSeek-R1 Collection: huggingface.co/collections/un…

Introducing Scandi-fine-web-cleaner, a decoder model trained to remove low-quality web from FineWeb 2 for Danish and Swedish - Uses FineWeb-c community annotations - 90%+ precision + minimal compute required - Enables efficient filtering of 43M+ documents

Annual reminder that you can stream datasets from HF during training - no need to download in advance :) (now with checkpointing + resuming too !)

Ross Wightman @wightmanr

2 years ago

So this is pretty wild, training with a ~1TB, ~12M sample image dataset, streaming directly from the HF hub over my fibre internet connection... works! Go go webdatset. Supported in timm via webdataset wrapper with curl pipes or @huggingface datasets streaming wrapper...

7 35 271 90K 110

🔍 Massive human feedback dataset for text-to-image models from @RapidataAI - 1.5M human responses from 152K participants - Evaluates image coherence, style & prompt alignment - Includes detailed error heatmaps - Covers DALL-E, Midjourney, Imagen outputs Available on @huggingface

We're excited to partner with the Mistral AI Hackathon: Game Jam Edition! 🕹️ 100+ participants, 48h to create AI-powered games on HF Spaces 👨‍🏫 @Jthmas404 and me will be there for mentoring + compute credits 📅 Jan 24-26, 2025, Paris ✍️ Apply by Jan 20 lu.ma/2rsuklms

Still catching up on ModernBERT after the holidays? (Its embedding model is already here!) On December 19th: A collaboration between Hugging Face, AnswerAI, and LightOn brought us ModernBERT. ModernBERT is a new encoder model family that outperforms BERT in many aspects while maintaining full backward compatibility: • Better: Outperforms on downstream tasks • Faster: 2-4x faster inference speeds • Longer: 16x larger sequence length On December 30th: Nomic AI brought us ModernBERT embed. ModernBERT embed is a new embedding model built on ModernBERT-base, with all its advances: • Better: Outperforms nomic-embed-text-v1 and nomic-embed-text-v1.5 on MTEB • Faster: Immediate FA2 and unpacking support for efficient inference • Longer: Embeds longer passages (up to 8,192 tokens) Both ModernBERT and ModernBERT embed are Apache 2.0 licensed (fully commercially permissible). The recipe notebook by my colleague Mary shows you how to use ModernBERT embed with @weaviate_io for vector search. (Fully open source. No registration and no API key are required) 🔗 Try it out: github.com/weaviate/recip…

Dolphin's getting frisky here! Another amazing dataset, @huggingface smoltalk, labeled for your filtering pleasure.

keyboard_arrow_left Previous keyboard_arrow_right Next

Reasoning datasets are flooding the top 30 trending datasets on Hugging Face. I created a collection of them here huggingface.co/collections/cf…

supercharge your LLM apps with smolagents 🔥 however cool your LLM is, without being agentic it can only go so far enter smolagents: a new agent library by @huggingface to make the LLM write code, do analysis and automate boring stuff!

our first release of 2025!

m_ric @AymericRoucher

a year ago

For months, we've worked on building @huggingface's new moonshot: agentic systems. So today we're very proud to announce the release of 𝚜𝚖𝚘𝚕𝚊𝚐𝚎𝚗𝚝𝚜! It's the simplest library we could make to let people build powerful agents. 💥 The main logic for agents fits in ~1000

47 263 2K 222K 2K

Fine-tuning ModernBERT for text classification using synthetic data generation From prompt to model in 3 steps. 1 simple datasets 20 minutes of generating 60 minutes of fine-tuning on my Macbook Pro Tutorial: buff.ly/4gwxBg7

🎅 Santa Claus has delivered the ultimate guide to understand OOM error (link in comment)

Stanford's compact 47-minute course on the essentials of the hugging face 🤗 transformers library and fine-tuning

Commits speak louder than words 🤪 * 4 new video models * Multiple image models, including SANA & Flux Control * New quantizers -> GGUF & TorchAO * New training scripts Enjoy 🤗

Massive Math dataset dropped on @huggingface 📐 FineMath consists of 34B tokens (FineMath-3+) and 54B tokens (FineMath-3+ with InfiMM-WebMath-3+) of mathematical educational content filtered from CommonCrawl. To curate this dataset, we trained a mathematical content classifier using annotations generated by LLama-3.1-70B-Instruct. We used the classifier to retain only the most educational mathematics content, focusing on clear explanations and step-by-step problem solving rather than advanced academic papers.