David Bau

papers

3,961

total citations

papers (16)

Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models

ICLR 2025arXiv

263

citations

Rewriting a Deep Generative Model

ECCV 2020arXiv

152

citations

Linearity of Relation Decoding in Transformer Language Models

ICLR 2024arXiv

143

citations

Diverse Image Generation via Self-Conditioned GANs

CVPR 2020arXiv

120

citations

Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking

ICLR 2024arXiv

citations

Editing a classifier by rewriting its prediction rules

NEURIPS 2021arXiv

citations

Sketch Your Own GAN

ICCV 2021arXiv

citations

FIND: A Function Description Benchmark for Evaluating Interpretability Methods

NEURIPS 2023arXiv

citations

Toward a Visual Concept Vocabulary for GAN Latent Space

ICCV 2021arXiv

citations

MIB: A Mechanistic Interpretability Benchmark

ICML 2025arXiv

citations

SliderSpace: Decomposing the Visual Capabilities of Diffusion Models

ICCV 2025arXiv

citations

When Are Concepts Erased From Diffusion Models?

NEURIPS 2025arXiv

citations

Disentangling Visual and Written Concepts in CLIP

CVPR 2022

citations

David Bau

papers (16)

Locating and Editing Factual Associations in GPT

Erasing Concepts from Diffusion Models

What makes fake images detectable? Understanding properties that generalize

Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models

Rewriting a Deep Generative Model

Linearity of Relation Decoding in Transformer Language Models

Diverse Image Generation via Self-Conditioned GANs

Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking

Editing a classifier by rewriting its prediction rules

Sketch Your Own GAN

FIND: A Function Description Benchmark for Evaluating Interpretability Methods

Toward a Visual Concept Vocabulary for GAN Latent Space

MIB: A Mechanistic Interpretability Benchmark

SliderSpace: Decomposing the Visual Capabilities of Diffusion Models

When Are Concepts Erased From Diffusion Models?

Disentangling Visual and Written Concepts in CLIP

papers (16)

Locating and Editing Factual Associations in GPT

Erasing Concepts from Diffusion Models

What makes fake images detectable? Understanding properties that generalize

Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models

Rewriting a Deep Generative Model

Linearity of Relation Decoding in Transformer Language Models

Diverse Image Generation via Self-Conditioned GANs

Fine-Tuning Enhances Existing Mechanisms: A Case Study on Entity Tracking

Editing a classifier by rewriting its prediction rules

Sketch Your Own GAN

FIND: A Function Description Benchmark for Evaluating Interpretability Methods

Toward a Visual Concept Vocabulary for GAN Latent Space

MIB: A Mechanistic Interpretability Benchmark

SliderSpace: Decomposing the Visual Capabilities of Diffusion Models

When Are Concepts Erased From Diffusion Models?

Disentangling Visual and Written Concepts in CLIP