lengquan88/dsh-dual-auto

README 摘要

dsh-dual-auto Dual-model auto-routing plugin for the DeepSeek Harness (dsh). Low-cost direct / high-cost upgrade with an escape-learning closed loop. Install Enable Add one row to your profile's cordis.patch.yml : Restart dsh web . The tools dual model route , dual model run , and dual model mark become available in every session. Tools Tool Purpose dual model route Six-criteria routing decision (length / context / domain coverage / rule conflict / confidence / novelty → six labels). Fingerprints that escaped once are force-upgraded. dual model run Decision + real model call: direct → deepseek-v4-flash , upgrade → deepseek-v4-pro (auto-degrade to flash on failure, marked degraded ). Probe tasks auto-validate against a gold set — wrong direct answers trigger escape learning. dual model mark Mark the quality of a direct result. correct=false learns the fingerprint and rewrites the disk log marker; the same fingerprint is force-upgraded next time. Persistence State persists to output/dsh router {fingerprints,stats}.json and dsh router decision log.jsonl — interoperable with the project's Python dao/model router.py (v2 dict fingerprints load directly). Links - npm: - Source mirror (ato…

分类