DSH Qwen38 Compaction Fix
zhubaohi/dsh-qwen38-compaction-fix · tools
DSH plugin: stop qwen3.8-27b from burning its output budget on thinking during context compaction.
★ 2GitHub stars
0Forks
2026-09-06Last updated
JavaScriptLanguage
MITLicense
Key features
- What id is being comparedthe model id your route serves, the one declared under llm-pi-ai.providers.
- That is the value dsh puts in the model field of every outgoing request to your gateway.
- It is not whatever name your gateway internally calls the model.
- Make it matcheither of two places works
- If nothing matches, the plugin silently does nothing: every request passes through unchanged, byte for byte, and there is no warning on a miss.
Requirements
- The wire fields this plugin rewrites are the ones the NInfer gateway interprets.
- Gateways served by other engines (llama.cpp, vLLM, FastMTP, ...) may ignore or spell these fields differently
Install command
dsh plugin --profile web add github:zhubaohi/dsh-qwen38-compaction-fix