Person
Researchers at Stanford University
2 stories · sorted newest first · 📡 RSS
Simpler Method DPO Outshines Others for Aligning Large Language Models
A recent study evaluates three methods for aligning LLMs without reinforcement learning, with DPO showing the highest MT Bench sco
New TimeScope Benchmark Evaluates Long Video Understanding Skills of VLMs
TimeScope, an open-source benchmark developed by Stanford University and Hugging Face, assesses localized retrieval, information s