1 paper · 1 filter
Yang Li, Zirui Zhang, Yang Liu +1
Current large language models reason in isolation. Although it is common to sample multiple reasoning paths in parallel, these trajectories do not interact, and often fail in the s…