phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-16 13:22:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge remote-tracking branch 'origin/main' into stack-config-default-embed

Browse source
This commit is contained in:
Ashwin Bharambe 2025-10-20 13:29:19 -07:00
commit 31249a1a75
237 changed files with 30895 additions and 15441 deletions
Download patch file Download diff file Expand all files Collapse all files

15
llama_stack/providers/remote/datasetio/huggingface/config.py
View file

@ -7,20 +7,17 @@ from typing import Any
from pydantic import BaseModel
from llama_stack.providers.utils.kvstore.config import (
KVStoreConfig,
SqliteKVStoreConfig,
)
from llama_stack.core.storage.datatypes import KVStoreReference
class HuggingfaceDatasetIOConfig(BaseModel):
kvstore: KVStoreConfig
kvstore: KVStoreReference
@classmethod
def sample_run_config(cls, __distro_dir__: str, **kwargs: Any) -> dict[str, Any]:
return {
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__,
db_name="huggingface_datasetio.db",
)
"kvstore": KVStoreReference(
backend="kv_default",
namespace="datasetio::huggingface",
).model_dump(exclude_none=True)
}

12
llama_stack/providers/remote/files/s3/config.py
View file

@ -8,7 +8,7 @@ from typing import Any
from pydantic import BaseModel, Field
from llama_stack.providers.utils.sqlstore.sqlstore import SqliteSqlStoreConfig, SqlStoreConfig
from llama_stack.core.storage.datatypes import SqlStoreReference
class S3FilesImplConfig(BaseModel):
@ -24,7 +24,7 @@ class S3FilesImplConfig(BaseModel):
auto_create_bucket: bool = Field(
default=False, description="Automatically create the S3 bucket if it doesn't exist"
)
metadata_store: SqlStoreConfig = Field(description="SQL store configuration for file metadata")
metadata_store: SqlStoreReference = Field(description="SQL store configuration for file metadata")
@classmethod
def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:
@ -35,8 +35,8 @@ class S3FilesImplConfig(BaseModel):
"aws_secret_access_key": "${env.AWS_SECRET_ACCESS_KEY:=}",
"endpoint_url": "${env.S3_ENDPOINT_URL:=}",
"auto_create_bucket": "${env.S3_AUTO_CREATE_BUCKET:=false}",
"metadata_store": SqliteSqlStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__,
db_name="s3_files_metadata.db",
),
"metadata_store": SqlStoreReference(
backend="sql_default",
table_name="s3_files_metadata",
).model_dump(exclude_none=True),
}

72
llama_stack/providers/remote/inference/nvidia/NVIDIA.md
View file

@ -45,7 +45,7 @@ The following example shows how to create a chat completion for an NVIDIA NIM.
```python
response = client.chat.completions.create(
model="meta-llama/Llama-3.1-8B-Instruct",
model="nvidia/meta/llama-3.1-8b-instruct",
messages=[
{
"role": "system",
@ -67,37 +67,40 @@ print(f"Response: {response.choices[0].message.content}")
The following example shows how to do tool calling for an NVIDIA NIM.
```python
from llama_stack.models.llama.datatypes import ToolDefinition, ToolParamDefinition
tool_definition = ToolDefinition(
tool_name="get_weather",
description="Get current weather information for a location",
parameters={
"location": ToolParamDefinition(
param_type="string",
description="The city and state, e.g. San Francisco, CA",
required=True,
),
"unit": ToolParamDefinition(
param_type="string",
description="Temperature unit (celsius or fahrenheit)",
required=False,
default="celsius",
),
tool_definition = {
"type": "function",
"function": {
"name": "get_weather",
"description": "Get current weather information for a location",
"parameters": {
"type": "object",
"properties": {
"location": {
"type": "string",
"description": "The city and state, e.g. San Francisco, CA",
},
"unit": {
"type": "string",
"description": "Temperature unit (celsius or fahrenheit)",
"default": "celsius",
},
},
"required": ["location"],
},
},
)
}
tool_response = client.chat.completions.create(
model="meta-llama/Llama-3.1-8B-Instruct",
model="nvidia/meta/llama-3.1-8b-instruct",
messages=[{"role": "user", "content": "What's the weather like in San Francisco?"}],
tools=[tool_definition],
)
print(f"Tool Response: {tool_response.choices[0].message.content}")
print(f"Response content: {tool_response.choices[0].message.content}")
if tool_response.choices[0].message.tool_calls:
for tool_call in tool_response.choices[0].message.tool_calls:
print(f"Tool Called: {tool_call.tool_name}")
print(f"Arguments: {tool_call.arguments}")
print(f"Tool Called: {tool_call.function.name}")
print(f"Arguments: {tool_call.function.arguments}")
```
### Structured Output Example
@ -105,33 +108,26 @@ if tool_response.choices[0].message.tool_calls:
The following example shows how to do structured output for an NVIDIA NIM.
```python
from llama_stack.apis.inference import JsonSchemaResponseFormat, ResponseFormatType
person_schema = {
"type": "object",
"properties": {
"name": {"type": "string"},
"age": {"type": "integer"},
"age": {"type": "number"},
"occupation": {"type": "string"},
},
"required": ["name", "age", "occupation"],
}
response_format = JsonSchemaResponseFormat(
type=ResponseFormatType.json_schema, json_schema=person_schema
)
structured_response = client.chat.completions.create(
model="meta-llama/Llama-3.1-8B-Instruct",
model="nvidia/meta/llama-3.1-8b-instruct",
messages=[
{
"role": "user",
"content": "Create a profile for a fictional person named Alice who is 30 years old and is a software engineer. ",
}
],
response_format=response_format,
extra_body={"nvext": {"guided_json": person_schema}},
)
print(f"Structured Response: {structured_response.choices[0].message.content}")
```
@ -141,7 +137,7 @@ The following example shows how to create embeddings for an NVIDIA NIM.
```python
response = client.embeddings.create(
model="nvidia/llama-3.2-nv-embedqa-1b-v2",
model="nvidia/nvidia/llama-3.2-nv-embedqa-1b-v2",
input=["What is the capital of France?"],
extra_body={"input_type": "query"},
)
@ -163,15 +159,15 @@ image_path = {path_to_the_image}
demo_image_b64 = load_image_as_base64(image_path)
vlm_response = client.chat.completions.create(
model="nvidia/vila",
model="nvidia/meta/llama-3.2-11b-vision-instruct",
messages=[
{
"role": "user",
"content": [
{
"type": "image",
"image": {
"data": demo_image_b64,
"type": "image_url",
"image_url": {
"url": f"data:image/png;base64,{demo_image_b64}",
},
},
{

9
llama_stack/providers/remote/inference/nvidia/nvidia.py
View file

@ -19,15 +19,6 @@ class NVIDIAInferenceAdapter(OpenAIMixin):
"""
NVIDIA Inference Adapter for Llama Stack.
Note: The inheritance order is important here. OpenAIMixin must come before
ModelRegistryHelper to ensure that OpenAIMixin.check_model_availability()
is used instead of ModelRegistryHelper.check_model_availability(). It also
must come before Inference to ensure that OpenAIMixin methods are available
in the Inference interface.
- OpenAIMixin.check_model_availability() queries the NVIDIA API to check if a model exists
- ModelRegistryHelper.check_model_availability() just returns False and shows a warning
"""
# source: https://docs.nvidia.com/nim/nemo-retriever/text-embedding/latest/support-matrix.html

2
llama_stack/providers/remote/vector_io/chroma/chroma.py
View file

@ -130,7 +130,7 @@ class ChromaVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtocolP
self.vector_db_store = None
async def initialize(self) -> None:
self.kvstore = await kvstore_impl(self.config.kvstore)
self.kvstore = await kvstore_impl(self.config.persistence)
self.vector_db_store = self.kvstore
if isinstance(self.config, RemoteChromaVectorIOConfig):

11
llama_stack/providers/remote/vector_io/chroma/config.py
View file

@ -8,20 +8,21 @@ from typing import Any
from pydantic import BaseModel, Field
from llama_stack.providers.utils.kvstore.config import KVStoreConfig, SqliteKVStoreConfig
from llama_stack.core.storage.datatypes import KVStoreReference
from llama_stack.schema_utils import json_schema_type
@json_schema_type
class ChromaVectorIOConfig(BaseModel):
url: str | None
kvstore: KVStoreConfig = Field(description="Config for KV store backend")
persistence: KVStoreReference = Field(description="Config for KV store backend")
@classmethod
def sample_run_config(cls, __distro_dir__: str, url: str = "${env.CHROMADB_URL}", **kwargs: Any) -> dict[str, Any]:
return {
"url": url,
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="chroma_remote_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::chroma_remote",
).model_dump(exclude_none=True),
}

11
llama_stack/providers/remote/vector_io/milvus/config.py
View file

@ -8,7 +8,7 @@ from typing import Any
from pydantic import BaseModel, ConfigDict, Field
from llama_stack.providers.utils.kvstore.config import KVStoreConfig, SqliteKVStoreConfig
from llama_stack.core.storage.datatypes import KVStoreReference
from llama_stack.schema_utils import json_schema_type
@ -17,7 +17,7 @@ class MilvusVectorIOConfig(BaseModel):
uri: str = Field(description="The URI of the Milvus server")
token: str | None = Field(description="The token of the Milvus server")
consistency_level: str = Field(description="The consistency level of the Milvus server", default="Strong")
kvstore: KVStoreConfig = Field(description="Config for KV store backend")
persistence: KVStoreReference = Field(description="Config for KV store backend")
# This configuration allows additional fields to be passed through to the underlying Milvus client.
# See the [Milvus](https://milvus.io/docs/install-overview.md) documentation for more details about Milvus in general.
@ -28,7 +28,8 @@ class MilvusVectorIOConfig(BaseModel):
return {
"uri": "${env.MILVUS_ENDPOINT}",
"token": "${env.MILVUS_TOKEN}",
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="milvus_remote_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::milvus_remote",
).model_dump(exclude_none=True),
}

2
llama_stack/providers/remote/vector_io/milvus/milvus.py
View file

@ -277,7 +277,7 @@ class MilvusVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtocolP
self.metadata_collection_name = "openai_vector_stores_metadata"
async def initialize(self) -> None:
self.kvstore = await kvstore_impl(self.config.kvstore)
self.kvstore = await kvstore_impl(self.config.persistence)
start_key = VECTOR_DBS_PREFIX
end_key = f"{VECTOR_DBS_PREFIX}\xff"
stored_vector_dbs = await self.kvstore.values_in_range(start_key, end_key)

13
llama_stack/providers/remote/vector_io/pgvector/config.py
View file

@ -8,7 +8,7 @@ from typing import Any
from pydantic import BaseModel, Field
from llama_stack.providers.utils.kvstore.config import KVStoreConfig, SqliteKVStoreConfig
from llama_stack.core.storage.datatypes import KVStoreReference
from llama_stack.schema_utils import json_schema_type
@ -19,7 +19,9 @@ class PGVectorVectorIOConfig(BaseModel):
db: str | None = Field(default="postgres")
user: str | None = Field(default="postgres")
password: str | None = Field(default="mysecretpassword")
kvstore: KVStoreConfig | None = Field(description="Config for KV store backend (SQLite only for now)", default=None)
persistence: KVStoreReference | None = Field(
description="Config for KV store backend (SQLite only for now)", default=None
)
@classmethod
def sample_run_config(
@ -38,7 +40,8 @@ class PGVectorVectorIOConfig(BaseModel):
"db": db,
"user": user,
"password": password,
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="pgvector_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::pgvector",
).model_dump(exclude_none=True),
}

2
llama_stack/providers/remote/vector_io/pgvector/pgvector.py
View file

@ -337,7 +337,7 @@ class PGVectorVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtoco
async def initialize(self) -> None:
log.info(f"Initializing PGVector memory adapter with config: {self.config}")
self.kvstore = await kvstore_impl(self.config.kvstore)
self.kvstore = await kvstore_impl(self.config.persistence)
await self.initialize_openai_vector_stores()
try:

11
llama_stack/providers/remote/vector_io/qdrant/config.py
View file

@ -8,7 +8,7 @@ from typing import Any
from pydantic import BaseModel
from llama_stack.providers.utils.kvstore.config import KVStoreConfig, SqliteKVStoreConfig
from llama_stack.core.storage.datatypes import KVStoreReference
from llama_stack.schema_utils import json_schema_type
@ -24,13 +24,14 @@ class QdrantVectorIOConfig(BaseModel):
prefix: str | None = None
timeout: int | None = None
host: str | None = None
kvstore: KVStoreConfig
persistence: KVStoreReference
@classmethod
def sample_run_config(cls, __distro_dir__: str, **kwargs: Any) -> dict[str, Any]:
return {
"api_key": "${env.QDRANT_API_KEY:=}",
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="qdrant_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::qdrant_remote",
).model_dump(exclude_none=True),
}

4
llama_stack/providers/remote/vector_io/qdrant/qdrant.py
View file

@ -161,9 +161,9 @@ class QdrantVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtocolP
self._qdrant_lock = asyncio.Lock()
async def initialize(self) -> None:
client_config = self.config.model_dump(exclude_none=True, exclude={"kvstore"})
client_config = self.config.model_dump(exclude_none=True, exclude={"persistence"})
self.client = AsyncQdrantClient(**client_config)
self.kvstore = await kvstore_impl(self.config.kvstore)
self.kvstore = await kvstore_impl(self.config.persistence)
start_key = VECTOR_DBS_PREFIX
end_key = f"{VECTOR_DBS_PREFIX}\xff"

13
llama_stack/providers/remote/vector_io/weaviate/config.py
View file

@ -8,7 +8,7 @@ from typing import Any
from pydantic import BaseModel, Field
from llama_stack.providers.utils.kvstore.config import KVStoreConfig, SqliteKVStoreConfig
from llama_stack.core.storage.datatypes import KVStoreReference
from llama_stack.schema_utils import json_schema_type
@ -16,14 +16,17 @@ from llama_stack.schema_utils import json_schema_type
class WeaviateVectorIOConfig(BaseModel):
weaviate_api_key: str | None = Field(description="The API key for the Weaviate instance", default=None)
weaviate_cluster_url: str | None = Field(description="The URL of the Weaviate cluster", default="localhost:8080")
kvstore: KVStoreConfig | None = Field(description="Config for KV store backend (SQLite only for now)", default=None)
persistence: KVStoreReference | None = Field(
description="Config for KV store backend (SQLite only for now)", default=None
)
@classmethod
def sample_run_config(cls, __distro_dir__: str, **kwargs: Any) -> dict[str, Any]:
return {
"weaviate_api_key": None,
"weaviate_cluster_url": "${env.WEAVIATE_CLUSTER_URL:=localhost:8080}",
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="weaviate_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::weaviate",
).model_dump(exclude_none=True),
}

4
llama_stack/providers/remote/vector_io/weaviate/weaviate.py
View file

@ -288,8 +288,8 @@ class WeaviateVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, NeedsRequestProv
async def initialize(self) -> None:
"""Set up KV store and load existing vector DBs and OpenAI vector stores."""
# Initialize KV store for metadata if configured
if self.config.kvstore is not None:
self.kvstore = await kvstore_impl(self.config.kvstore)
if self.config.persistence is not None:
self.kvstore = await kvstore_impl(self.config.persistence)
else:
self.kvstore = None
log.info("No kvstore configured, registry will not persist across restarts")