Picture for Wieland Brendel

Wieland Brendel

VGGSounder: Audio-Visual Evaluations for Foundation Models

Add code
Aug 12, 2025
Viaarxiv icon

Estimating Treatment Effects with Independent Component Analysis

Add code
Jul 22, 2025
Viaarxiv icon

An Empirically Grounded Identifiability Theory Will Accelerate Self-Supervised Learning Research

Add code
Apr 17, 2025
Viaarxiv icon

LLMs on the Line: Data Determines Loss-to-Loss Scaling Laws

Add code
Feb 17, 2025
Viaarxiv icon

Interaction Asymmetry: A General Principle for Learning Composable Abstractions

Add code
Nov 12, 2024
Viaarxiv icon

Cross-Entropy Is All You Need To Invert the Data Generating Process

Add code
Oct 29, 2024
Figure 1 for Cross-Entropy Is All You Need To Invert the Data Generating Process
Figure 2 for Cross-Entropy Is All You Need To Invert the Data Generating Process
Figure 3 for Cross-Entropy Is All You Need To Invert the Data Generating Process
Figure 4 for Cross-Entropy Is All You Need To Invert the Data Generating Process
Viaarxiv icon

In Search of Forgotten Domain Generalization

Add code
Oct 10, 2024
Viaarxiv icon

InfoNCE: Identifying the Gap Between Theory and Practice

Add code
Jun 28, 2024
Viaarxiv icon

Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning

Add code
Jun 20, 2024
Figure 1 for Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning
Figure 2 for Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning
Figure 3 for Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning
Figure 4 for Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning
Viaarxiv icon

Understanding LLMs Requires More Than Statistical Generalization

Add code
May 03, 2024
Viaarxiv icon