Tag: adversarial reinforcement learning
Browse our exclusive articles!
Token-Space Attacks on Reward Models in RLHF
Discover how token-space attacks exploit reward models in RLHF, revealing vulnerabilities beyond semantic manipulation and impacting AI safety.
Limits of Reinforcement Learning Alignment in AI Safety
Explore the generalization limits of reinforcement learning alignment and its impact on AI safety in large language models with compound jailbreaks analysi...
Moondream Segmentation: Advanced Image Masking AI
Discover Moondream Segmentation, an AI model enhancing image masks from verbal cues with cutting-edge reinforcement learning and autoregressive decoding.
PRISM: Interpretable Policy Reuse in Reinforcement Learning
Discover PRISM, a framework for interpretable policy reuse in reinforcement learning enabling zero-shot transfer across diverse agents.
OPRIDE: Efficient Offline Preference-Based Reinforcement Learning
Discover OPRIDE, a novel algorithm improving offline preference-based reinforcement learning with efficient in-dataset exploration and reduced human feedba...