Jonas Geiping

Affiliations

University of Maryland, College Park

papers

4,534

total citations

papers (20)

Inverting Gradients - How easy is it to break privacy in federated learning?

NEURIPS 2020arXiv

1,528

citations

Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models

CVPR 2023arXiv

435

citations

Universal Guidance for Diffusion Models

ICLR 2024arXiv

399

citations

Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery

NEURIPS 2023arXiv

370

citations

An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks

ICML 2025arXiv

citations

Can Language Models Falsify? Evaluating Algorithmic Reasoning with Counterexample Creation

COLM 2025arXiv

citations

Tree-Rings Watermarks: Invisible Fingerprints for Diffusion Images

NEURIPS 2023

citations

Jonas Geiping

Affiliations

papers (20)

Inverting Gradients - How easy is it to break privacy in federated learning?

Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models

Universal Guidance for Diffusion Models

Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery

Cold Diffusion: Inverting Arbitrary Image Transforms Without Noise

Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Text

MetaPoison: Practical General-purpose Clean-label Data Poisoning

Understanding and Mitigating Copying in Diffusion Models

On the Reliability of Watermarks for Large Language Models

Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach

Adversarial Examples Make Strong Poisons

On the Exploitability of Instruction Tuning

Autoregressive Perturbations for Data Poisoning

A Performance-Driven Benchmark for Feature Selection in Tabular Deep Learning

What Can We Learn from Unlearnable Datasets?

Great Models Think Alike and this Undermines AI Oversight

Object Recognition as Next Token Prediction

An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks

Can Language Models Falsify? Evaluating Algorithmic Reasoning with Counterexample Creation

Tree-Rings Watermarks: Invisible Fingerprints for Diffusion Images

papers (20)

Inverting Gradients - How easy is it to break privacy in federated learning?

Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Models

Universal Guidance for Diffusion Models

Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery

Cold Diffusion: Inverting Arbitrary Image Transforms Without Noise

Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Text

MetaPoison: Practical General-purpose Clean-label Data Poisoning

Understanding and Mitigating Copying in Diffusion Models

On the Reliability of Watermarks for Large Language Models

Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach

Adversarial Examples Make Strong Poisons

On the Exploitability of Instruction Tuning

Autoregressive Perturbations for Data Poisoning

A Performance-Driven Benchmark for Feature Selection in Tabular Deep Learning

What Can We Learn from Unlearnable Datasets?

Great Models Think Alike and this Undermines AI Oversight

Object Recognition as Next Token Prediction

An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks

Can Language Models Falsify? Evaluating Algorithmic Reasoning with Counterexample Creation

Tree-Rings Watermarks: Invisible Fingerprints for Diffusion Images