AI RESEARCH
SSPO: Subsentence-level Policy Optimization
arXiv CS.CL
•
ArXi:2511.04256v2 Announce Type: replace As a key component of large language model (LLM) post-