Publications

Selected papers. The full list, about 5,000 citations as a named author, is on Google Scholar.

• 2026LogAct: Enabling Agentic Reliability via Shared LogsBalakrishnan, Bharambe, Testuggine, et al. arXiv.
• 2024Introducing v0.5 of the AI Safety Benchmark from MLCommonsVidgen et al. Meta representative on the working group. arXiv.
• 2024The Llama 3 Herd of ModelsLlama Team. Contributor. arXiv.
• 2023Llama Guard: LLM-based Input-Output Safeguard for Human-AI ConversationsInan, Upasani, Chi, …, Testuggine, Khabsa. arXiv.
• 2022Are Multimodal Transformers Robust to Missing Modality?Ma, Ren, Zhao, Testuggine, Peng. CVPR.
• 2021Opacus: User-Friendly Differential Privacy Library in PyTorchYousefpour, Shilov, Sablayrolles, Testuggine, et al. NeurIPS PriML workshop.
• 2020The Hateful Memes Challenge: Detecting Hate Speech in Multimodal MemesKiela, Firooz, Mohan, Goswami, Singh, Ringshia, Testuggine. NeurIPS.
• 2019Supervised Multimodal Bitransformers for Classifying Images and TextKiela, Bhooshan, Firooz, Testuggine. NeurIPS workshops.