1 11 21

Matthias De Paolis

llmat

https://mattdepaolis.github.io/blog/

AI & ML interests

Post-training, model editing

Recent Activity

liked a model 19 days ago

deepseek-ai/DeepSeek-V4-Flash

liked a model 2 months ago

nvidia/Qwen3.5-397B-A17B-NVFP4

upvoted an article 2 months ago

Mixture of Experts (MoEs) in Transformers

View all activity

Organizations

liked a model 19 days ago

deepseek-ai/DeepSeek-V4-Flash

Text Generation • 158B • Updated 7 days ago • 1.37M • • 1.07k

liked a model 2 months ago

nvidia/Qwen3.5-397B-A17B-NVFP4

Text Generation • Updated Mar 30 • 744k • 98

liked a Space 3 months ago

Reachy Mini

🤖

140

All about Reachy Mini, from building to getting started

liked a model 3 months ago

Dream-org/Dream-v0-Instruct-7B

Text Generation • 8B • Updated Jul 15, 2025 • 181k • 156

liked 2 Spaces 3 months ago

Unlocking On-Policy Distillation for Any Model Family

📝

102

Visualize on-policy distillation for any model family

Evaluation Guidebook

📝

317

Explore LLM benchmark trends over time

liked 2 models 3 months ago

Qwen/Qwen3-Coder-Next-FP8

Text Generation • 80B • Updated Feb 3 • 355k • • 144

zai-org/GLM-5

Text Generation • 754B • Updated Apr 5 • 190k • • 2.09k

liked a model 6 months ago

baidu/ERNIE-4.5-VL-28B-A3B-Thinking

Image-Text-to-Text • 30B • Updated Mar 6 • 2.2k • 539

liked a Space 6 months ago

The Smol Training Playbook

📚

3.17k

The secrets to building world-class LLMs

liked a model about 1 year ago

deepseek-ai/DeepSeek-R1

Text Generation • 685B • Updated Mar 27, 2025 • 3.83M • • 13.3k

liked a Space about 1 year ago

Open LLM Leaderboard

🏆

14k

Track, rank and evaluate open LLMs and chatbots

liked 4 models over 1 year ago

liked 4 models almost 2 years ago

llmat/donut-invoices-v1

Image-Text-to-Text • Updated Sep 19, 2023 • 6 • 1

llmat/germanBert-invoices-v1

Text Classification • Updated Jul 8, 2024 • 5 • 1

llmat/Mistral-v0.3-7B-ORPO

Text Generation • 7B • Updated Aug 5, 2024 • 35 • 1

mistralai/Mistral-Large-Instruct-2407

Updated Jul 28, 2025 • 7.41k • 860