mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-24 05:45:11 +00:00
bca69f4439
The three routes were served without a response schema, so /openapi.json published an empty schema for their 200 bodies and a client generating types from the document got unknown for each ModelGroupInfoResponse and ModelInfoV1Response, both named ResourceResponse envelopes, now back the route decorators. /model_group/info returns the model, so FastAPI serializes it. /model/info keeps its orjson Response, so there the model documents the body: a list of deployments, or the single deployment object the proxy serves when started from the CLI with --model, whose litellm_params carry every unset non-secret value as the string None, which is why the params are a wire-specific model rather than the router's input type The dashboard's generated schema.d.ts is regenerated with npm run gen:api |
||
|---|---|---|
| .. | ||
| litellm-dashboard | ||
| Dockerfile | ||
| nginx.conf | ||