Fine-grained LLM verification for DSH agents: pairwise rewards over score-token logprobs, Probabilistic Pivot Tournament best-of-N selection, and per-step progress tracking.
View on GitHubBrowse similar plugins
Fine-grained LLM verification for DSH agents: pairwise rewards over score-token logprobs, Probabilistic Pivot Tournament best-of-N selection, and per-step progress tracking.
See the GitHub repository for installation instructions — this plugin does not publish a standard install command.
中文说明:Fine-grained LLM verification for DSH agents: pairwise rewards over score-token logprobs, Probabilistic Pivot Tournament best-of-N selection, and per-step progress tracking.(安装方式请参考 GitHub 仓库说明。)