Niels Rogge (@NielsRogge) / X

Niels Rogge

3,391 posts

Niels Rogge

@NielsRogge

ML Engineer @huggingface. Building paperswithco.de. @KU_Leuven grad. General interest in machine & deep learning. Making AI more accessible for everyone!

Belgium

Joined April 2010

Pinned
Niels Rogge
@NielsRogge
Sep 29, 2022
Today my Transformers-Tutorials repo hit 2,000 stars on @github! 🤩 Very greatful :) the repo contains many tutorial notebooks on inference + fine-tuning with custom data for Transformers on all kinds of data; text, images, scanned PDFs, videos ⭐
GitHub - NielsRogge/Transformers-Tutorials: This repository contains demos I made with the Transf...
From github.com
Niels Rogge
@NielsRogge
Jan 21, 2025
“So there’s this Chinese company called DeepSeek which basically does what OpenAI initially intended to do. They open-sourced a model trained with large-scale reinforcement learning, beating everyone else, and even releasing a paper detailing their process“
212K
Niels Rogge
@NielsRogge
Sep 18, 2025
The most legendary LLM release is still Mistral-7B for me No context given, just a magnet link First time we got a decent model running locally Feels like yesterday
121K
Niels Rogge
@NielsRogge
Jan 7, 2025
Wait how big of an AI lab is @deepseek_ai ?
252K
Niels Rogge
@NielsRogge
Oct 24, 2024
Microsoft silently dropped a new model on the hub 👀 "OmniParser is a general screen parsing tool, which interprets/converts UI screenshot to structured format, to improve existing LLM based UI agents"
microsoft/OmniParser · Hugging Face
From huggingface.co
171K
Niels Rogge
@NielsRogge
Oct 18, 2025
Karpathy: "RL is terrible" Every RL researcher on the Karpathy interview: "I agree with everything he says"
180K
Niels Rogge
@NielsRogge
Jun 19, 2025
"Hugging Face is basically the equivalent of Github in the era of software 2.0" - Karpathy, 2025, colorized
85K
Niels Rogge
@NielsRogge
Jul 10, 2025
I feel like people who know about distributed training, CUDA kernels, JAX/Flax, GPU clusters, profiling etc have a massive edge currently Maybe I should learn those too
130K
Niels Rogge
@NielsRogge
Jan 29, 2025
Absolutely disgusting post! Literally the opposite of what @huggingface stands for
Dario Amodei
@DarioAmodei
Jan 29, 2025
My thoughts on China, export controls and two possible futures darioamodei.com/on-deepseek-an…
143K
Niels Rogge
@NielsRogge
Mar 16, 2025
I learned how to migrate to uv: 1) uv init 2) uv venv 3) source .venv/bin/activate 4) uv add -r requirements.txt Congrats, you can now delete your requirements.txt file
Han
@HanchungLee
Mar 15, 2025
uv still has a bunch of issues. - it’s not pip - lack of ide and coding agent support - venv created by uv cant be used without uv - poor documentation. there’s no playbook on how to migrate to it. information scattered around like a changelog.
187K
Niels Rogge
@NielsRogge
Aug 5, 2022
OWL-ViT by @GoogleAI is now available @huggingface Transformers. The model is a minimal extension of CLIP for zero-shot object detection given text queries. 🤯 🥳 It has impressive generalization capabilities and is a great first step for open-vocabulary object detection! (1/2)
GIF
Niels Rogge
@NielsRogge
Mar 18, 2025
Amazing video on multi-head attention, multi-query attention, grouped-query attention and MLA (multi-latent attention), the main innovation of DeepSeek's V3 model Made with @3blue1brown's amazing Manim library
59K
Niels Rogge
@NielsRogge
May 29, 2025
When people ask me: "Which framework should I use for developing AI agents? Should I use @OpenAI Agent SDK? Google's Agent Development Kit? LangGraph or @pydantic Agent framework?" Here's my honest answer:
132K
Niels Rogge
@NielsRogge
Dec 6, 2024
Feeling the same tbh
82K