Pinjia He

GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

2023 · 14 Zit. · arXiv (Cornell University)

LogicAsker: Evaluating and Improving the Logical Reasoning Ability of Large Language Models

2024 · 2 Zit. · arXiv (Cornell University)

Rigor, Reliability, and Reproducibility Matter: A Decade-Scale Survey of 572 Code Benchmarks

2025 · 0 Zit. · ArXiv.org