Somesh Jha

papers

233

total citations

papers (11)

AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs

ICLR 2025arXiv

115

citations

Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles

NEURIPS 2021arXiv

citations

A Quantitative Geometric Approach to Neural-Network Smoothness

NEURIPS 2022arXiv

citations

Do Large Code Models Understand Programming Concepts? Counterfactual Analysis for Code Predicates

ICML 2024arXiv

citations

Two Heads are Actually Better than One: Towards Better Adversarial Robustness via Transduction and Rejection

ICML 2024arXiv

citations

Grounding Neural Inference with Satisfiability Modulo Theories

NEURIPS 2023

citations

Robust Learning against Relational Adversaries

NEURIPS 2022

citations

Somesh Jha

papers (11)

AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs

Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles

A Quantitative Geometric Approach to Neural-Network Smoothness

Do Large Code Models Understand Programming Concepts? Counterfactual Analysis for Code Predicates

Robust and Actively Secure Serverless Collaborative Learning

Overparameterization from Computational Constraints

A Separation Result Between Data-oblivious and Data-aware Poisoning Attacks

Validating Mechanistic Interpretations: An Axiomatic Approach

Two Heads are Actually Better than One: Towards Better Adversarial Robustness via Transduction and Rejection

Grounding Neural Inference with Satisfiability Modulo Theories

Robust Learning against Relational Adversaries

papers (11)

AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs

Detecting Errors and Estimating Accuracy on Unlabeled Data with Self-training Ensembles

A Quantitative Geometric Approach to Neural-Network Smoothness

Do Large Code Models Understand Programming Concepts? Counterfactual Analysis for Code Predicates

Robust and Actively Secure Serverless Collaborative Learning

Overparameterization from Computational Constraints

A Separation Result Between Data-oblivious and Data-aware Poisoning Attacks

Validating Mechanistic Interpretations: An Axiomatic Approach

Two Heads are Actually Better than One: Towards Better Adversarial Robustness via Transduction and Rejection

Grounding Neural Inference with Satisfiability Modulo Theories

Robust Learning against Relational Adversaries