Hugging Face's logo
Activity Feed Request to join this org

AI & ML interests

None defined yet.

Recent Activity
Lukas431ย  authored a paper 16 days ago
Self-Supervised Learning of Structured Dynamics from Videos
Lukas431ย  submitted a paper 16 days ago
Self-Supervised Learning of Structured Dynamics from Videos
NoeFlandreย  authored a paper 20 days ago
Composing Verifiable Conceptual Models via Building Blocks: Towards Design-Time Verification of Agentic AI Workflows
View all activity
eienmojikiย 
posted an update about 1 month ago
lbourdoisย 
posted an update 2 months ago
view post
Post
1208
New blog post!
An introduction to a little-known but highly effective model reduction method: ๐—ง๐—ฟ๐—ถ๐—บ๐—บ๐—ถ๐—ป๐—ดโœ‚๏ธ
We show how to reduce model size (we went up to 87.24% reduction) while preserving its performance.

We applied this technique to 16 different model families across several modalities to illustrate that it works on any architecture (as long as the embedding layer is the last one of the model) and on any modality involving text.
From these 16 families, we generated over ๐Ÿฑ,๐Ÿฑ๐Ÿฌ๐Ÿฌ ๐—บ๐—ผ๐—ป๐—ผ๐—น๐—ถ๐—ป๐—ด๐˜‚๐—ฎ๐—น ๐—บ๐—ผ๐—ฑ๐—ฒ๐—น๐˜€ ๐—ถ๐—ป ๐Ÿญ๐Ÿฎ๐Ÿฐ ๐—ฑ๐—ถ๐—ณ๐—ณ๐—ฒ๐—ฟ๐—ฒ๐—ป๐˜ ๐—น๐—ฎ๐—ป๐—ด๐˜‚๐—ฎ๐—ด๐—ฒ๐˜€ ๐ŸŒ

Key takeaways from our experiments:
1๏ธโƒฃ Trimming does not require a GPU. Our models were obtained on a CPU.
2๏ธโƒฃ This method scales up to at least 4B parameters (we did not test beyond that).
3๏ธโƒฃ Trimmed model is smaller than the original while preserving its performance. If you observe a slight performance drop, just fine-tuned to recover or even surpass the original performance.
4๏ธโƒฃ For an equivalent compute budget, it is better to trim then fine-tune rather than fine-tuning the original model. Since the model is smaller, you can run more epochs/show more data and get in fine a better model than the original.
5๏ธโƒฃ Trimming is a competitive alternative to distillation and quantization. E.g. we obtained our alternative to DistilBERT in 9 minutes on CPU vs. 90 hours of GPU for the latter.
6๏ธโƒฃ Trimming could generate reasoning traces in the language of the trimmed model. This could be an alternative to generating traces in English and then translating them into the desired language.

And many other things (such as how much data are needed, the impact of the database used, the order in which it should be done, etc.) are available in the blogpost!

Blogpost: https://huggingface.co/blog/lbourdois/introduction-to-trimming
Models: alphaedge-ai/Trimming_models_search
  • 4 replies
ยท
alexicanesseย 
submitted a paper to Daily Papers 3 months ago
alexicanesseย 
authored 4 papers 3 months ago
alceballosaย 
authored a paper 6 months ago
Jofthomasย 
posted an update 8 months ago
view post
Post
4733
The new Mistral 3 models are here !

Today, we announce Mistral 3, the next generation of Mistral models. Mistral 3 includes three state-of-the-art small, dense models (14B, 8B, and 3B) and Mistral Large 3 โ€“ our most capable model to date โ€“ a sparse mixture-of-experts trained with 41B active and 675B total parameters.

All models are released under the Apache 2.0 license.

Ministrals :
https://huggingface.co/collections/mistralai/ministral-3

Mistral Large 3:
https://huggingface.co/collections/mistralai/mistral-large-3
  • 2 replies
ยท
lbourdoisย 
posted an update 10 months ago
gregmialzย 
authored a paper 11 months ago
clemย 
posted an update about 1 year ago
clemย 
posted an update about 1 year ago
clemย 
posted an update about 1 year ago
view post
Post
8127
Today, we're unveiling two new open-source AI robots! HopeJR for $3,000 & Reachy Mini for $300 ๐Ÿค–๐Ÿค–๐Ÿค–

Let's go open-source AI robotics!
  • 7 replies
ยท
clemย 
posted an update about 1 year ago
view post
Post
3986
It's just become easier to share your apps on the biggest AI app store (aka HF spaces) for unlimited storage, more visibility and community interactions.

Just pick a React, Svelte, or Vue template when you create your space or add app_build_command: npm run build in your README's YAML and app_file: build/index.html in your README's YAML block.

Or follow this link: https://huggingface.co/new-space?sdk=static

Let's build!
  • 1 reply
ยท
clemย 
posted an update about 1 year ago
view post
Post
4181
Playing with Veo3 this morning. Share your prompt if you want me to create videos for you (bonus point if they funnily reference HF/open-source). These videos are "a cat on the moon rapping "I love Hugging Face""!
  • 27 replies
ยท
Jofthomasย 
posted an update about 1 year ago
view post
Post
4767
Meet our new agentic model : ๐——๐—ฒ๐˜ƒ๐˜€๐˜๐—ฟ๐—ฎ๐—น

Devstral is an open-source LLM built software engineering tasks built under a collaboration between Mistral AI and All Hands AI ๐Ÿ™Œ.

๐—ž๐—ฒ๐˜† ๐—ณ๐—ฒ๐—ฎ๐˜๐˜‚๐—ฟ๐—ฒ๐˜€ :
โ€ข ๐Ÿค– ๐—”๐—ด๐—ฒ๐—ป๐˜๐˜€ : perfect for Agentic coding
โ€ข ๐Ÿƒ ๐—น๐—ถ๐—ด๐—ต๐˜๐˜„๐—ฒ๐—ถ๐—ด๐—ต๐˜: Devstral is a ๐Ÿฎ๐Ÿฐ๐—• parameter based on Mistral small.
โ€ข ยฉ๏ธ ๐—”๐—ฝ๐—ฎ๐—ฐ๐—ต๐—ฒ ๐Ÿฎ.๐Ÿฌ, meaning fully open-source !
โ€ข ๐Ÿ“„ A ๐Ÿญ๐Ÿฎ๐Ÿด๐—ธ context window.

๐Ÿ“šBlog : https://mistral.ai/news/devstral
โšกAPI : The model is also available on our API under the name ๐—ฑ๐—ฒ๐˜ƒ๐˜€๐˜๐—ฟ๐—ฎ๐—น-๐˜€๐—บ๐—ฎ๐—น๐—น-๐Ÿฎ๐Ÿฑ๐Ÿฌ๐Ÿฑ
๐Ÿค— repo : mistralai/Devstral-Small-2505

Can't wait to see what you will build with it !
  • 1 reply
ยท
clemย 
posted an update about 1 year ago
view post
Post
3278
Very cool to see
pytorch contributing on Hugging Face. Time to follow them to see what they're cooking!
  • 2 replies
ยท
clemย 
posted an update over 1 year ago
clemย 
posted an update over 1 year ago
view post
Post
4142
What are you using to evaluate models or AI systems? So far we're building lighteval & leaderboards on the hub but still feels early & a lot more to build. What would be useful to you?
  • 6 replies
ยท