Bootstrapping LLM Robustness for VLM Safety via Reducing the Pretraining Modality Gap

Documenti analoghi

• LoRA is All You Need for Safety Alignment of Reasoning LLMs
  di: Xue, Yihao, et al.
  Pubblicazione: (2025)
• Better Safe than Sorry: Pre-training CLIP against Targeted Data Poisoning and Backdoor Attacks
  di: Yang, Wenhan, et al.
  Pubblicazione: (2023)
• Tuning the Implicit Regularizer of Masked Diffusion Language Models: Enhancing Generalization via Insights from $k$-Parity
  di: Huang, Jianhao, et al.
  Pubblicazione: (2026)
• Data-Efficient Contrastive Self-supervised Learning: Most Beneficial Examples for Supervised Learning Contribute the Least
  di: Joshi, Siddharth, et al.
  Pubblicazione: (2023)
• Beyond Semantic Entropy: Boosting LLM Uncertainty Quantification with Pairwise Semantic Similarity
  di: Nguyen, Dang, et al.
  Pubblicazione: (2025)