A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
Asifullah Khan, Muhammad Zaeem Khan, Saleha Jamshed, et al.
A comprehensive survey of LLM advances covering reasoning, adaptability, efficiency, and ethics, highlighting techniques like Chain-of-Thought, instruction tuning, and RLHF, and identifying future research directions.
Shunyu Liu, Wenkai Fang, Zetian Hu, et al.
This survey systematically reviews Direct Preference Optimization (DPO), a streamlined alternative to RLHF for aligning LLMs with human preferences, covering its variants, applications, and future directions.
Unknown
This paper extends the analysis of length exploitation from classical RLHF to Direct Preference Optimization (DPO), studying how DPO can over-optimize for response length rather than quality.
Unknown
Direct Preference Optimization (DPO) simplifies RLHF by directly optimizing language model policies from preference data, eliminating the need for a separate reward model.
Unknown
This paper provides a comprehensive overview of RLHF, covering its core methods, challenges, and applications in building large-scale machine learning systems.
Unknown
This paper introduces variational preference learning (VPL), a method that personalizes RLHF by learning a latent user-specific preference distribution from diverse human feedback.
Josef Dai, Xuehai Pan, Ruiyang Sun, et al.
Safe RLHF decouples reward modeling and safety constraints to train language models that are both helpful and safe.
Harrison Lee, Samrat Phatale, H. Mansoor, et al.
This paper introduces RLAIF, a method that uses AI feedback from a large language model to scale reinforcement learning from human feedback, achieving comparable performance to RLHF.
Unknown
This survey provides a comprehensive overview of reinforcement learning from human feedback (RLHF), covering its variants, applications, and open challenges.
Unknown
This paper surveys open problems and fundamental limitations of RLHF, arguing that current methods are brittle and may not scale to reliably align AI systems with human values.
Unknown
This paper provides a critical analysis of RLHF, examining its methodologies, challenges, and implications for aligning large language models with human values.
Venkata Naga Sai Vishnu Rohit Pulipaka, Anish Katta, Deva Rohit Reddy Peddireddy
This paper shows that a native C++ ONNX Runtime inference engine can score RLHF rewards faster than PyTorch eager mode and FastAPI on CPU, but torch.compile is faster on GPU, and batching strategy matters more than language or runtime choice.