Paper 'DecodingTrust' received the Outstanding Paper Award at NeurIPS 2023 and the 2024 Cybersecurity Award for Best Machine Learning and Security Paper
1 paper accepted to ICML 2025
4 papers accepted to ICLR 2025, including 'MMDT', 'ChatQA 2', 'AdvAgent', 'EIA', and 'AdvWave'
2 papers accepted to AAAI 2025
1 paper accepted to CVPR 2024
Released MMDT in September 2024: a comprehensive benchmark for trustworthiness in multimodal foundation models
Released ChatQA 2 in September 2024: a Llama 3.0-based model with enhanced long-context understanding and RAG capabilities
Released UltraLong-8B in April 2025: models supporting up to 4M context length with competitive performance on standard tasks