phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2026-01-02 21:10:01 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'main' into sambanova-inferene

Browse source
This commit is contained in:
snova-edwardm 2025-01-14 10:04:52 -08:00 • committed by GitHub
commit 89ab2be302
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
385 changed files with 39001 additions and 9280 deletions
Download patch file Download diff file Expand all files Collapse all files

27
llama_stack/providers/registry/inference.py
View file

@ -6,8 +6,13 @@
from typing import List
from llama_stack.distribution.datatypes import * # noqa: F403
from llama_stack.providers.datatypes import (
AdapterSpec,
Api,
InlineProviderSpec,
ProviderSpec,
remote_provider_spec,
)
META_REFERENCE_DEPS = [
"accelerate",
@ -18,6 +23,7 @@ META_REFERENCE_DEPS = [
"transformers",
"zmq",
"lm-format-enforcer",
"sentence-transformers",
]
@ -52,6 +58,13 @@ def available_providers() -> List[ProviderSpec]:
module="llama_stack.providers.inline.inference.vllm",
config_class="llama_stack.providers.inline.inference.vllm.VLLMConfig",
),
InlineProviderSpec(
api=Api.inference,
provider_type="inline::sentence-transformers",
pip_packages=["sentence-transformers"],
module="llama_stack.providers.inline.inference.sentence_transformers",
config_class="llama_stack.providers.inline.inference.sentence_transformers.config.SentenceTransformersInferenceConfig",
),
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
@ -141,6 +154,16 @@ def available_providers() -> List[ProviderSpec]:
provider_data_validator="llama_stack.providers.remote.inference.together.TogetherProviderDataValidator",
),
),
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_type="groq",
pip_packages=["groq"],
module="llama_stack.providers.remote.inference.groq",
config_class="llama_stack.providers.remote.inference.groq.GroqConfig",
provider_data_validator="llama_stack.providers.remote.inference.groq.GroqProviderDataValidator",
),
),
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(