Thomas Wolf's picture

Thomas Wolf PRO

thomwolf

·

https://thomwolf.io

AI & ML interests

NLP and open-source :-)

Recent Activity

updated a Space about 5 hours ago

science/README

reacted to lewtun's post with 🔥 about 21 hours ago

I was initially pretty sceptical about Meta's Coconut paper [1] because the largest perf gains were reported on toy linguistic problems. However, these results on machine translation are pretty impressive! https://x.com/casper_hansen_/status/1875872309996855343 Together with the recent PRIME method [2] for scaling RL, reasoning for open models is looking pretty exciting for 2025! [1] https://huggingface.co/papers/2412.06769 [2] https://huggingface.co/blog/ganqu/prime

liked a model 1 day ago

deepseek-ai/DeepSeek-V3

View all activity

Articles

Introducing smolagents: simple agents that write actions in code.

FineWeb2-C: Help Build Better Language Models in Your Language

LeMaterial: an open source initiative to accelerate materials discovery and research

FineVideo: behind the scenes

Fine-tuning LLMs to 1.58bit: extreme quantization made easy

A failed experiment: Infini-Attention, and why we should keep trying?

Jack of All Trades, Master of Some, a Multi-Purpose Transformer Agent

Constitutional AI with Open LLMs

Open LLM Leaderboard: DROP deep dive

What's going on with the Open LLM Leaderboard?

Can foundation models label data like humans?

Organizations

thomwolf's activity

updated a Space about 5 hours ago

README

reacted to lewtun's post with 🔥 about 21 hours ago

Post

1474

I was initially pretty sceptical about Meta's Coconut paper [1] because the largest perf gains were reported on toy linguistic problems. However, these results on machine translation are pretty impressive!

https://x.com/casper_hansen_/status/1875872309996855343

Together with the recent PRIME method [2] for scaling RL, reasoning for open models is looking pretty exciting for 2025!

[1] Training Large Language Models to Reason in a Continuous Latent Space (2412.06769)
[2] https://huggingface.co/blog/ganqu/prime

liked a model 1 day ago

deepseek-ai/DeepSeek-V3

Updated 8 days ago • 71.7k • 1.32k

upvoted an article 1 day ago

Article

🐺🐦‍⬛ LLM Comparison/Test: DeepSeek-V3, QVQ-72B-Preview, Falcon3 10B, Llama 3.3 70B, Nemotron 70B in my updated MMLU-Pro CS benchmark

By

•

4 days ago

• 30

upvoted a collection 1 day ago

Phi-3

Phi-3 family of small language and multi-modal models. Language models are available in short- and long-context lengths. • 26 items • Updated Nov 14, 2024 • 543

liked a Space 1 day ago

AI Phone Leaderboard

AI Phone Leaderboard

liked a model 1 day ago

matteogeniaccio/phi-4

Updated 24 days ago • 46.5k • 181

upvoted a paper 1 day ago

Phi-4 Technical Report

Paper • 2412.08905 • Published 26 days ago • 97

upvoted a paper 2 days ago

Aguvis: Unified Pure Vision Agents for Autonomous GUI Interaction

Paper • 2412.04454 • Published Dec 5, 2024 • 57

updated a Space 3 days ago

Discussion Forum

liked 2 models 4 days ago

Qwen/Qwen2.5-3B-Instruct

Text Generation • Updated Sep 25, 2024 • 556k • • 130

PowerInfer/SmallThinker-3B-Preview

Text Generation • Updated about 12 hours ago • 6.37k • 262

liked a Space 5 days ago

Get Travel Duration Tool

liked a dataset 11 days ago

HuggingFaceTB/finemath

Viewer • Updated 14 days ago • 48.3M • 31.9k • 226

upvoted a collection 12 days ago

Falcon3

Falcon3 family of Open Foundation Models is a set of pretrained and instruct LLMs ranging from 1B to 10B parameters. • 40 items • Updated 18 days ago • 75

liked a model 12 days ago

deepseek-ai/DeepSeek-V3-Base

Updated 8 days ago • 8.36k • 1.16k

upvoted an article 14 days ago

Article

FineWeb2-C: Help Build Better Language Models in Your Language

By

•

14 days ago

• 11

liked a Space 16 days ago

ECCV24 Papers

liked a model 16 days ago

IamCreateAI/Ruyi-Mini-7B

Image-to-Video • Updated 12 days ago • 16k • 565

liked a dataset 16 days ago

data-is-better-together/fineweb-c

Updated about 5 hours ago • 643 • 32