Tencent’s new AI technique teaches language models ‘parallel thinking’

The Parallel-R1 framework uses reinforcement learning to teach models how to explore multiple reasoning paths at once, ...