4 papers
RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback
Harrison Lee, Samrat Phatale, Hassan Mansoor +8
Reinforcement learning from human feedback (RLHF) has proven effective in aligning large language models (LLMs) with human preferences, but gathering high-quality preference labels…
ScreenAI: A Vision-Language Model for UI and Infographics Understanding
Gilles Baechler, Srinivas Sunkara, Maria Wang +7
Screen user interfaces (UIs) and infographics, sharing similar visual language and design principles, play important roles in human communication and human-machine interaction. We…
LLMs cannot find reasoning errors, but can correct them given the error location
Gladys Tyen, Hassan Mansoor, Victor CÄrbune +2
While self-correction has shown promise in improving LLM outputs in terms of style and quality (e.g. Chen et al., 2023b; Madaan et al., 2023), recent attempts to self-correct logic…
VQA Training Sets are Self-play Environments for Generating Few-shot Pools
Tautvydas Misiunas, Hassan Mansoor, Jasper Uijlings +2
Large-language models and large-vision models are increasingly capable of solving compositional reasoning tasks, as measured by breakthroughs in visual-question answering benchmark…