Thanks, Mikyo. Passing the api_base in model_kwargs worked. Appreciate it!
While I have you, do you know how I can pass authentication parameters to a secured endpoint? The api_base is a model served and secured by a token.
response = litellm.completion(
model="ollama/llama2",
messages=messages,
api_base="https://mydomain.com/serving-endpoints/llama2-7b-chat-hf/invocations"
)