| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
feat: harden memory model compatibility Add typed OpenAI-compatible provider controls for thinking, temperature, output token parameter, structured output, reasoning-only correction, JSON repair, and per-stage token budgets for ram-a-mem ingestion. Update ARM64 GLM deployment assets, verification scripts, and RAM-A memory model compatibility documentation. Refs: openeuler/RAM-A#6 Issue: https://atomgit.com/openeuler/RAM-A/issues/6 | 1 个月前 |