API & Parameters¶
Authoritative reference for the OpenAI-compatible API, UCM configuration keys, environment variables, and UCM-relevant engine parameters. This page is a placeholder — populate it per the spec below.
What to add¶
Reader goal: look up a parameter's default, valid values, and type when running UCM.
Required content (the single source of truth for all parameters, maintained as structured data):
- OpenAI-Compatible API: endpoints, request/response schema, UCM extension fields.
- UCM configuration keys (
ucm_config_example.yaml): for each key — name / type / default / valid values / one-line meaning / which store or engine uses it. - Environment variables: same fields as above (
ENABLE_UCM_PATCH,PLATFORM,MOONCAKE_CONFIG_PATH,HICACHE_ENGINE,UCM_CONFIG_FILE,ASCEND_RT_VISIBLE_DEVICES). - Engine parameters: vLLM / SGLang / MindIE parameters related to UCM / KV cache integration (not the engine's full parameter set).
Don't:
- Don't write task tutorials (that's User Guide's job).
- Don't repeat the parameter subsets already inlined in User Guide task pages; this page is the authoritative table only.
Acceptance:
- Covers every parameter already used across task pages.
- Each entry has type / default / valid values.
- No conflicts with task pages.
First-draft approach: extract parameters already in use from
model-tour/*, capabilities/prefix-cache/*, and quick_start/* into a table,
then back-validate task-page consistency.
Owner: (to be assigned)
Reference¶
- Compatibility & Metrics — supported models and metrics.
- Prefix Cache stores — backend config semantics.
- GLM-5.1 4-node PD — a worked
ucm_config_example.yaml+vllm serveexample.