enable streaming support, use openai-python instead of httpx

This commit is contained in:
Matthew Farrellee 2024-11-04 10:22:29 -05:00
parent 2dd8c4bcb6
commit dbe665ed19
7 changed files with 1037 additions and 341 deletions

View file

@ -144,7 +144,9 @@ def available_providers() -> List[ProviderSpec]:
api=Api.inference,
adapter=AdapterSpec(
adapter_type="nvidia",
pip_packages=[], # TODO(mf): need to specify httpx if it's already a llama-stack dep?
pip_packages=[
"openai",
],
module="llama_stack.providers.adapters.inference.nvidia",
config_class="llama_stack.providers.adapters.inference.nvidia.NVIDIAConfig",
),