Tag: Direct Preference Optimization
Browse our exclusive articles!
LifeAlign: Lifelong Alignment for Large Language Models
Discover LifeAlign, a novel framework enhancing lifelong alignment in large language models with memory-augmented preference optimization.
Why Random Sampling Beats Active Selection in Modern LLMs
Discover why random sampling outperforms active selection in online DPO for modern LLMs, highlighting efficiency and performance insights.
VERTIGO: Optimizing Cinematic Camera Trajectories Visually
Discover VERTIGO, a framework that enhances cinematic camera trajectories by optimizing visual preferences using AI and real-time rendering.
Reducing LLM Biases with Debiasing Direct Preference Optimization
Discover how Debiasing-DPO reduces LLM biases from spurious social contexts, improving accuracy and fairness in high-stakes decision-making.
Mitigating Contextual Exposure Bias in Speech-LLMs
Discover effective strategies to reduce contextual exposure bias in Speech-LLMs, enhancing ASR accuracy and robustness in noisy real-world environments.