Personer

Francesco Croce

ELLIS Institute

Assistant Professor

T313 Dept. Computer Science

Full researcher profile

https://research.aalto.fi/...

E-post

francesco.croce@aalto.fi

Forskningsgrupp

Computer Science Professors, Assistant Professor
Professorship Croce Francesco, Assistant Professor

Publikationer

Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks

Maksym Andriushchenko, Francesco Croce, Nicolas Flammarion 2025 13th International Conference on Learning Representations, ICLR 2025

Adversarially Robust CLIP Models Can Induce Better (Robust) Perceptual Metrics

Francesco Croce, Christian Schlarmann, Naman Deep Singh, Matthias Hein 2025 Proceedings - 2025 IEEE Conference on Secure and Trustworthy Machine Learning, SaTML 2025

Selective induction Heads: How Transformers Select Causal Structures in Context

Francesco D'Angelo, Francesco Croce, Nicolas Flammarion 2025 13th International Conference on Learning Representations, ICLR 2025

Is In-Context Learning Sufficient for Instruction Following in LLMs?

Hao Zhao, Maksym Andriushchenko, Francesco Croce, Nicolas Flammarion 2025 13th International Conference on Learning Representations, ICLR 2025

JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models

Patrick Chao, Edoardo Debenedetti, Alexander Robey, Maksym Andriushchenko, Francesco Croce, Vikash Sehwag, Edgar Dobriban, Nicolas Flammarion, George J. Pappas, Florian Tramèr, Hamed Hassani, Eric Wong 2024 Advances in Neural Information Processing Systems 37 (NeurIPS 2024)

Francesco Croce

Forskningsgrupp

Publikationer

Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks

Adversarially Robust CLIP Models Can Induce Better (Robust) Perceptual Metrics

Selective induction Heads: How Transformers Select Causal Structures in Context

Is In-Context Learning Sufficient for Instruction Following in LLMs?

JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models

Evaluating and Improving the Robustness of Image Classifiers against Adversarial Attacks

Segment (Almost) Nothing: Prompt-Agnostic Adversarial Attacks on Segmentation Models

Towards Reliable Evaluation and Fast Training of Robust Semantic Segmentation Models

Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models

Long Is More for Alignment: A Simple but Tough-to-Beat Baseline for Instruction Fine-Tuning