mirror of
https://github.com/BerriAI/litellm.git
synced 2026-11-16 09:22:22 +00:00
Router defaults now fill only the keys a deployment leaves unset, so a deployment's own value (vertex_location, api_base, drop_params) beats default_litellm_params. The proxy token counters, drop-params checks, and the Entra cooldown check use the same order |
||
|---|---|---|
| .. | ||
| __init__.py | ||
| test_call.py | ||
| test_main.py | ||