iiinova — where your digital self keeps evolving

Luisa Zintgraf

@luisazintgraf.bsky.social

2.6K followers 129 following 11 posts

RL & Meta-Learning @ DeepMind.

Posts Replies Media

Luisa Zintgraf @luisazintgraf.bsky.social · 06/11/2025

Huge shout-out to my co-first authors @dancalian.bsky.social, @gregfar.bsky.social, & Iurii Kemaev. And to our amazing collaborators: Matteo Hessel, Jeremy Shar, Junhyuk Oh, András György, @schaul.bsky.social, @jeffdean.bsky.social, Hado van Hasselt, & Dave Silver.

150

Luisa Zintgraf @luisazintgraf.bsky.social · 06/11/2025

We believe that the DataRater is a promising step towards more automated and principled dataset curation. This could be especially important for filtering and making the best use of massive synthetic datasets in the future. For a deeper dive, check out arxiv.org/pdf/2505.17895

160

Luisa Zintgraf @luisazintgraf.bsky.social · 06/11/2025

So what does the DataRater learn? It automatically identifies and down-weights data that aligns with human intuitions of low quality, such as incorrect text encodings, OCR errors, and irrelevant content.

150

Luisa Zintgraf @luisazintgraf.bsky.social · 06/11/2025

The result? The DataRater is highly effective at filtering data, leading to significant compute efficiency improvements. In our experiments, we observed up to a 46.6% net compute gain while often improving final model performance.

121

Luisa Zintgraf @luisazintgraf.bsky.social · 06/11/2025

We introduce the DataRater, a meta-learning method that learns to rate the value of each data point for training. Instead of manually specifying filtering rules, we train the DataRater to optimize for a simple goal: improving the training efficiency on a held-out dataset.

140

Luisa Zintgraf @luisazintgraf.bsky.social · 06/11/2025

Foundation models are trained on large datasets, but not all data is created equal. Dataset curation often relies on manual, coarse-grained filtering and hand-crafted rules. This is becoming a major challenge, especially with the rise of synthetic data.

140

Luisa Zintgraf @luisazintgraf.bsky.social · 06/11/2025

Excited to share our new paper, "DataRater: Meta-Learned Dataset Curation"! We explore a fundamental question: How can we *automatically* learn which data is most valuable for training foundation models? Paper: arxiv.org/pdf/2505.17895 to appear at @neuripsconf.bsky.social Thread 👇

1264

Luisa Zintgraf @luisazintgraf.bsky.social · 09/04/2025

Tagging first author @jakeabeck.bsky.social who just joined bsky! Welcome 🎉

010

Luisa Zintgraf @luisazintgraf.bsky.social · 09/04/2025

📘 Journal: nowpublishers.com/article/Deta... 📝 ArXiv: arxiv.org/abs/2301.08028 🎙️ Podcast: www.talkrl.com/episodes/jac... 🎥 Talk: youtu.be/XUQ9jLOZqGc

youtu.be

[AUTOML23] A Tutorial on MetaReinforcement Learning

YouTube video by AutoMLConf

161

Luisa Zintgraf @luisazintgraf.bsky.social · 09/04/2025

🎉 Our Meta-RL survey is now published in Foundations and Trends in Machine Learning! A deep dive into how agents can learn to learn 🤖🧠 Huge kudos to Jake Beck & Risto Vuorio for leading the charge, and to co-authors Evan Liu, Zheng Xiong, Chelsea Finn & @shimon8282.bsky.social!

1303

Luisa Zintgraf @luisazintgraf.bsky.social · 05/02/2025

Wanna work on Gemini? DeepMind is hiring! 🚀

040

Reposted by Luisa Zintgraf

Joost van Amersfoort @joost.bsky.social · 05/02/2025

Interested in helping us make Gemini Pro even better? The Gemini pre-training team is looking for a Research Scientist in London to push the boundaries of LLM scaling: understanding, predicting, and improving. ♊️🚀 Apply here: boards.greenhouse.io/deepmind/job...

boards.greenhouse.io

Research Scientist, Large Scale Pre-Training Model

London, UK

085