5 papers
PolyAlign: Conditional Human-Distribution Alignment
L. D. M. S. Sai Teja, Ufaq Khan, Sathira Silva +2
Post-training methods such as supervised fine-tuning (SFT) and preference optimization typically align language models toward a single global assistant behavior. While effective fo…
Diverging Preferences: When do Annotators Disagree and do Models Know?
Michael JQ Zhang, Zhilin Wang, Jena D. Hwang +6
We examine diverging preferences in human-labeled preference datasets. We develop a taxonomy of disagreement sources spanning ten categories across four high-level classes and find…
HelpSteer3-Preference: Open Human-Annotated Preference Data across Diverse Tasks and Languages
Zhilin Wang, Jiaqi Zeng, Olivier Delalleau +6
Preference datasets are essential for training general-domain, instruction-following language models with Reinforcement Learning from Human Feedback (RLHF). Each subsequent data re…
HelpSteer3: Human-Annotated Feedback and Edit Data to Empower Inference-Time Scaling in Open-Ended General-Domain Tasks
Zhilin Wang, Jiaqi Zeng, Olivier Delalleau +6
Inference-Time Scaling has been critical to the success of recent models such as OpenAI o1 and DeepSeek R1. However, many techniques used to train models for inference-time scaling…
HelpSteer2-Preference: Complementing Ratings with Preferences
Zhilin Wang, Alexander Bukharin, Olivier Delalleau +5
Reward models are critical for aligning models to follow instructions, and are typically trained following one of two popular paradigms: Bradley-Terry style or Regression style. Ho…