2.1k citations · 5.7k across the 61 of their papers we have counts for
65 papers
When Do Learned Blocks Preserve Causal Meaning? Block Identifiability and Constructive Causal Abstractions
Karim Zaghw, Wieland Brendel, Patrik Reizinger
Standard identifiability theory assumes one-dimensional latent factors, but learned representations can encode latent information across multidimensional blocks whose internal coor…
LAION-BVD: A 10-Million-Hour Open Video Dataset for Multimodal Pre-training
Andreas Hochlehnert, Marianna Nezhurina, Mehdi Cherti +9
We present LAION-BVD, a large-scale open video dataset for multimodal learning, which contains 1.3B platform-specific video URLs collected from CommonCrawl. From these, we download…
LittleLearner: Language Models Under Pedagogically Controlled Knowledge Exposure
Fanfei Li, Jana Zeller, Manuel Prada-Corral +4
Modern language models are trained on heterogeneous web-scale text corpora. Consequently, studying knowledge and skill acquisition is difficult, as prior exposure to related conten…
HALLMARK: How to Diagnose LLM Citation Verifiers and When to Deploy Them
Patrik Reizinger, Kayoon Kim, Wieland Brendel
GPTZero found 53 NeurIPS 2025 accepted papers with hallucinated citations, and venues now screen submissions with automated reference checkers. Rule- and Large Language Model (LLM)…
Stabilizing Extrapolation in Looped Transformers via Learned Stochastic Stopping
Hsun-Yu Kuo, El Mahdi Chayti, Patrik Reizinger +2
Looped Transformers, which repeatedly apply a shared transformer block, are an architecturally natural fit for variable-length algorithmic tasks. Although they can exhibit strong l…
Causality is Key for Interpretability Claims to Generalise
Shruti Joshi, Aaron Mueller, David Klindt +3
Interpretability research on large language models (LLMs) has yielded important insights into model behaviour, yet recurring pitfalls persist: findings that do not generalise, and…
Is Generation Required for Data-Efficient Perception?
Jack Brady, Bernhard Schölkopf, Thomas Kipf +2
It has been hypothesized that achieving the data efficiency of human visual perception requires a generative approach in which internal representations result from inverting a deco…
MATH-Beyond: A Benchmark for RL to Expand Beyond the Base Model
Prasanna Mayilvahanan, Ricardo Dominguez-Olmedo, Thaddäus Wiedemer +1
With the advent of DeepSeek-R1, a new wave of reinforcement learning (RL) methods has emerged that seem to unlock stronger mathematical reasoning. However, a closer look at the ope…
VGGSounder: Audio-Visual Evaluations for Foundation Models
Daniil Zverev, Thaddäus Wiedemer, Ameya Prabhu +3
The emergence of audio-visual foundation models underscores the importance of reliably assessing their multi-modal understanding. The VGGSound dataset is commonly used as a benchma…
Skill Learning via Policy Diversity Yields Identifiable Representations for Reinforcement Learning
Patrik Reizinger, Bálint Mucsányi, Siyuan Guo +3
Self-supervised feature learning and pretraining methods in reinforcement learning (RL) often rely on information-theoretic principles, termed mutual information skill learning (MI…
Estimating Treatment Effects with Independent Component Analysis
Patrik Reizinger, Lester Mackey, Wieland Brendel +1
Independent Component Analysis (ICA) uses a measure of non-Gaussianity to identify latent sources from data and estimate their mixing coefficients (Shimizu et al., 2006). Meanwhile…
Position: An Empirically Grounded Identifiability Theory Will Accelerate Self-Supervised Learning Research
Patrik Reizinger, Randall Balestriero, David Klindt +1
Self-Supervised Learning (SSL) powers many current AI systems. As research interest and investment grow, the SSL design space continues to expand. The Platonic view of SSL, followi…
Pretraining Frequency Predicts Compositional Generalization of CLIP on Real-World Tasks
Thaddäus Wiedemer, Yash Sharma, Ameya Prabhu +2
We investigate the success conditions for compositional generalization of CLIP models on real-world data through performance prediction. Prior work shows that CLIP requires exponen…
LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws
Prasanna Mayilvahanan, Thaddäus Wiedemer, Sayak Mallick +2
Scaling laws guide the development of large language models (LLMs) by offering estimates for the optimal balance of model size, tokens, and compute. More recently, loss-to-loss sca…
Interaction Asymmetry: A General Principle for Learning Composable Abstractions
Jack Brady, Julius von Kügelgen, Sébastien Lachapelle +3
Learning disentangled representations of concepts and re-composing them in unseen ways is crucial for generalizing to out-of-domain situations. However, the underlying properties o…
Rule Extrapolation in Language Models: A Study of Compositional Generalization on OOD Prompts
Anna Mészáros, Szilvia Ujváry, Wieland Brendel +2
LLMs show remarkable emergent abilities, such as inferring concepts from presumably out-of-distribution prompts, known as in-context learning. Though this success is often attribut…
Cross-Entropy Is All You Need To Invert the Data Generating Process
Patrik Reizinger, Alice Bizeul, Attila Juhos +4
Supervised learning has become a cornerstone of modern machine learning, yet a comprehensive theory explaining its effectiveness remains elusive. Empirical phenomena, such as neura…
In Search of Forgotten Domain Generalization
Prasanna Mayilvahanan, Roland S. Zimmermann, Thaddäus Wiedemer +4
Out-of-Domain (OOD) generalization is the ability of a model trained on one or more domains to generalize to unseen domains. In the ImageNet era of computer vision, evaluation sets…
InfoNCE: Identifying the Gap Between Theory and Practice
Evgenia Rusak, Patrik Reizinger, Attila Juhos +3
Prior theory work on Contrastive Learning via the InfoNCE loss showed that, under certain assumptions, the learned representations recover the ground-truth latent factors. We argue…
Position: Understanding LLMs Requires More Than Statistical Generalization
Patrik Reizinger, Szilvia Ujváry, Anna Mészáros +3
The last decade has seen blossoming research in deep learning theory attempting to answer, "Why does deep learning generalize?" A powerful shift in perspective precipitated this pr…
Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning
Patrik Reizinger, Siyuan Guo, Ferenc Huszár +2
Identifying latent representations or causal structures is important for good generalization and downstream task performance. However, both fields have been developed rather indepe…
Effective pruning of web-scale datasets based on complexity of concept clusters
Amro Abbas, Evgenia Rusak, Kushal Tirumala +3
Utilizing massive web-scale datasets has led to unprecedented performance gains in machine learning models, but also imposes outlandish compute requirements for their training. In…
An Interventional Perspective on Identifiability in Gaussian LTI Systems with Independent Component Analysis
Goutham Rajendran, Patrik Reizinger, Wieland Brendel +1
We investigate the relationship between system identification and intervention design in dynamical systems. While previous research demonstrated how identifiable representation lea…
Does CLIP's Generalization Performance Mainly Stem from High Train-Test Similarity?
Prasanna Mayilvahanan, Thaddäus Wiedemer, Evgenia Rusak +2
Foundation models like CLIP are trained on hundreds of millions of samples and effortlessly generalize to new tasks and inputs. Out of the box, CLIP shows stellar zero-shot and few…