Can Models Keep Getting Stronger After Deployment? SERPO Enables Label‑Free Self‑Evolution on Open‑Ended Tasks

The SERPO framework replaces answer‑voting with a self‑evolving rubric, allowing test‑time reinforcement learning to improve large language models on open‑ended tasks without any human labels, and demonstrates up to 73% of privileged‑supervision gains on benchmarks such as HealthBench and ResearchQA.

AI evaluationHealthBenchOpen-Ended Tasks

0 likes · 15 min read

Can Models Keep Getting Stronger After Deployment? SERPO Enables Label‑Free Self‑Evolution on Open‑Ended Tasks

Machine Heart

Aug 5, 2026 · Artificial Intelligence

Can Large Language Models Self‑Evolve Beyond Math and Code?

The article introduces RLSVR, a reinforcement‑learning framework that creates self‑verifiable rewards for open‑ended tasks via task transformation, and its SpyRL implementation, showing substantial gains on summarization, creative writing, and math benchmarks without relying on external reward models.

Large Language ModelsOpen-Ended TasksRLSVR

0 likes · 13 min read

Can Large Language Models Self‑Evolve Beyond Math and Code?