Paul Setinek retweeted
What do more than 7000 training runs tell us about the data scaling behavior of neural surrogates?
In our #NeurIPS2026 paper, we strategically harnessed the cost-accuracy trade-off of numerical simulators to reduce data-generation compute by up to 6.3x.
More details in 🧵
I am very happy to share the first paper of my PhD:
SIMSHIFT - A Benchmark for Adapting Neural Surrogates to Distribution Shifts
1/n
We introduce SIMSHIFT: A Benchmark for Adapting Neural Surrogates to Distribution Shifts. I sincerely hope that new ideas are coming out from this benchmark.
Paper: arxiv.org/abs/2506.12007
Code: github.com/psetinek/simshift
📄 Paper: arxiv.org/abs/2506.12007
💻 Code: github.com/psetinek/simshift
🤗 Dataset: huggingface.co/datasets/sims…
5/n
1
32
A huge thanks goes out to my amazing collaborators @ggalletti_ , @jo_brandstetter and @wzell7 . Additionally, many thanks to Thomas Gross and Dominik Schnürer from Linz Center of Mechatronics GmbH for their support throughout this project.
6/n
54
Paul Setinek retweeted
I am excited to present the result of a fruitful internship at @AIatMeta. We introduce preference discerning, which denotes in-context conditioning on user preferences expressed in text to steer the recommendation system. This enhances flexibility and personalization. 1/n
Newly published research for generative retrieval for recommendations from teams at Meta.
- Preference Discerning with LLM-Enhanced Generative Retrieval ➡️ go.fb.me/evvcu8
- Unifying Generative and Dense Retrieval for Sequential Recommendation ➡️ go.fb.me/i7l955
Paul Setinek retweeted
Relying on this formula to measure predictive uncertainty? You might measure the wrong thing, depending on your assumptions. Time to shed light on the basics of uncertainty estimation. 🧵👇
Paul Setinek retweeted
Interesting in scaling up neural operators? Happy to announce that Universal Physics Transformers (UPT) -- a scalable framework for neural operators is accepted at #neurips2024.
Paper: arxiv.org/abs/2402.12365
Project page: ml-jku.github.io/UPT/













