phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-14 13:02:36 +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

4
llama_stack/providers/inline/agents/meta_reference/agents.py
View file

@ -83,8 +83,8 @@ class MetaReferenceAgentsImpl(Agents):
self.policy = policy
async def initialize(self) -> None:
self.persistence_store = await kvstore_impl(self.config.persistence_store)
self.responses_store = ResponsesStore(self.config.responses_store, self.policy)
self.persistence_store = await kvstore_impl(self.config.persistence.agent_state)
self.responses_store = ResponsesStore(self.config.persistence.responses, self.policy)
await self.responses_store.initialize()
self.openai_responses_impl = OpenAIResponsesImpl(
inference_api=self.inference_api,

32
llama_stack/providers/inline/agents/meta_reference/config.py
View file

@ -8,24 +8,30 @@ from typing import Any
from pydantic import BaseModel
from llama_stack.providers.utils.kvstore import KVStoreConfig
from llama_stack.providers.utils.kvstore.config import SqliteKVStoreConfig
from llama_stack.providers.utils.sqlstore.sqlstore import SqliteSqlStoreConfig, SqlStoreConfig
from llama_stack.core.storage.datatypes import KVStoreReference, ResponsesStoreReference
class AgentPersistenceConfig(BaseModel):
"""Nested persistence configuration for agents."""
agent_state: KVStoreReference
responses: ResponsesStoreReference
class MetaReferenceAgentsImplConfig(BaseModel):
persistence_store: KVStoreConfig
responses_store: SqlStoreConfig
persistence: AgentPersistenceConfig
@classmethod
def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:
return {
"persistence_store": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__,
db_name="agents_store.db",
),
"responses_store": SqliteSqlStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__,
db_name="responses_store.db",
),
"persistence": {
"agent_state": KVStoreReference(
backend="kv_default",
namespace="agents",
).model_dump(exclude_none=True),
"responses": ResponsesStoreReference(
backend="sql_default",
table_name="responses",
).model_dump(exclude_none=True),
}
}

1
llama_stack/providers/inline/agents/meta_reference/responses/openai_responses.py
View file

@ -359,6 +359,7 @@ class OpenAIResponsesImpl:
tool_executor=self.tool_executor,
safety_api=self.safety_api,
guardrail_ids=guardrail_ids,
instructions=instructions,
)
# Stream the response

4
llama_stack/providers/inline/agents/meta_reference/responses/streaming.py
View file

@ -110,6 +110,7 @@ class StreamingResponseOrchestrator:
text: OpenAIResponseText,
max_infer_iters: int,
tool_executor, # Will be the tool execution logic from the main class
instructions: str,
safety_api,
guardrail_ids: list[str] | None = None,
):
@ -133,6 +134,8 @@ class StreamingResponseOrchestrator:
self.accumulated_usage: OpenAIResponseUsage | None = None
# Track if we've sent a refusal response
self.violation_detected = False
# system message that is inserted into the model's context
self.instructions = instructions
async def _create_refusal_response(self, violation_message: str) -> OpenAIResponseObjectStream:
"""Create a refusal response to replace streaming content."""
@ -176,6 +179,7 @@ class StreamingResponseOrchestrator:
tools=self.ctx.available_tools(),
error=error,
usage=self.accumulated_usage,
instructions=self.instructions,
)
async def create_response(self) -> AsyncIterator[OpenAIResponseObjectStream]:

12
llama_stack/providers/inline/batches/reference/config.py
View file

@ -6,13 +6,13 @@
from pydantic import BaseModel, Field
from llama_stack.providers.utils.kvstore.config import KVStoreConfig, SqliteKVStoreConfig
from llama_stack.core.storage.datatypes import KVStoreReference
class ReferenceBatchesImplConfig(BaseModel):
"""Configuration for the Reference Batches implementation."""
kvstore: KVStoreConfig = Field(
kvstore: KVStoreReference = Field(
description="Configuration for the key-value store backend.",
)
@ -33,8 +33,8 @@ class ReferenceBatchesImplConfig(BaseModel):
@classmethod
def sample_run_config(cls, __distro_dir__: str) -> dict:
return {
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__,
db_name="batches.db",
),
"kvstore": KVStoreReference(
backend="kv_default",
namespace="batches",
).model_dump(exclude_none=True),
}

15
llama_stack/providers/inline/datasetio/localfs/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 LocalFSDatasetIOConfig(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="localfs_datasetio.db",
)
"kvstore": KVStoreReference(
backend="kv_default",
namespace="datasetio::localfs",
).model_dump(exclude_none=True)
}

15
llama_stack/providers/inline/eval/meta_reference/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 MetaReferenceEvalConfig(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="meta_reference_eval.db",
)
"kvstore": KVStoreReference(
backend="kv_default",
namespace="eval",
).model_dump(exclude_none=True)
}

12
llama_stack/providers/inline/files/localfs/config.py
View file

@ -8,14 +8,14 @@ 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 LocalfsFilesImplConfig(BaseModel):
storage_dir: str = Field(
description="Directory to store uploaded files",
)
metadata_store: SqlStoreConfig = Field(
metadata_store: SqlStoreReference = Field(
description="SQL store configuration for file metadata",
)
ttl_secs: int = 365 * 24 * 60 * 60 # 1 year
@ -24,8 +24,8 @@ class LocalfsFilesImplConfig(BaseModel):
def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:
return {
"storage_dir": "${env.FILES_STORAGE_DIR:=" + __distro_dir__ + "/files}",
"metadata_store": SqliteSqlStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__,
db_name="files_metadata.db",
),
"metadata_store": SqlStoreReference(
backend="sql_default",
table_name="files_metadata",
).model_dump(exclude_none=True),
}

33
llama_stack/providers/inline/telemetry/meta_reference/telemetry.py
View file

@ -4,6 +4,7 @@
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
import os
import threading
from typing import Any
@ -60,26 +61,28 @@ class TelemetryAdapter(Telemetry):
# Recreating the telemetry adapter multiple times will result in duplicate span processors.
# Since the library client can be recreated multiple times in a notebook,
# the kernel will hold on to the span processor and cause duplicate spans to be written.
if _TRACER_PROVIDER is None:
provider = TracerProvider()
trace.set_tracer_provider(provider)
_TRACER_PROVIDER = provider
if os.environ.get("OTEL_EXPORTER_OTLP_ENDPOINT"):
if _TRACER_PROVIDER is None:
provider = TracerProvider()
trace.set_tracer_provider(provider)
_TRACER_PROVIDER = provider
# Use single OTLP endpoint for all telemetry signals
# Use single OTLP endpoint for all telemetry signals
# Let OpenTelemetry SDK handle endpoint construction automatically
# The SDK will read OTEL_EXPORTER_OTLP_ENDPOINT and construct appropriate URLs
# https://opentelemetry.io/docs/languages/sdk-configuration/otlp-exporter
span_exporter = OTLPSpanExporter()
span_processor = BatchSpanProcessor(span_exporter)
trace.get_tracer_provider().add_span_processor(span_processor)
# Let OpenTelemetry SDK handle endpoint construction automatically
# The SDK will read OTEL_EXPORTER_OTLP_ENDPOINT and construct appropriate URLs
# https://opentelemetry.io/docs/languages/sdk-configuration/otlp-exporter
span_exporter = OTLPSpanExporter()
span_processor = BatchSpanProcessor(span_exporter)
trace.get_tracer_provider().add_span_processor(span_processor)
metric_reader = PeriodicExportingMetricReader(OTLPMetricExporter())
metric_provider = MeterProvider(metric_readers=[metric_reader])
metrics.set_meter_provider(metric_provider)
metric_reader = PeriodicExportingMetricReader(OTLPMetricExporter())
metric_provider = MeterProvider(metric_readers=[metric_reader])
metrics.set_meter_provider(metric_provider)
else:
logger.warning("OTEL_EXPORTER_OTLP_ENDPOINT is not set, skipping telemetry")
self.meter = metrics.get_meter(__name__)
self._lock = _global_lock
async def initialize(self) -> None:

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

@ -8,14 +8,14 @@ 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):
db_path: str
kvstore: KVStoreConfig = Field(description="Config for KV store backend")
persistence: KVStoreReference = Field(description="Config for KV store backend")
@classmethod
def sample_run_config(
@ -23,7 +23,8 @@ class ChromaVectorIOConfig(BaseModel):
) -> dict[str, Any]:
return {
"db_path": db_path,
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="chroma_inline_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::chroma",
).model_dump(exclude_none=True),
}

9
llama_stack/providers/inline/vector_io/faiss/config.py
View file

@ -8,16 +8,19 @@ 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
@json_schema_type
class FaissVectorIOConfig(BaseModel):
kvstore: KVStoreConfig
persistence: 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="faiss_store.db")
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::faiss",
).model_dump(exclude_none=True)
}

2
llama_stack/providers/inline/vector_io/faiss/faiss.py
View file

@ -184,7 +184,7 @@ class FaissVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtocolPr
self.cache: dict[str, VectorDBWithIndex] = {}
async def initialize(self) -> None:
self.kvstore = await kvstore_impl(self.config.kvstore)
self.kvstore = await kvstore_impl(self.config.persistence)
# Load existing banks from kvstore
start_key = VECTOR_DBS_PREFIX
end_key = f"{VECTOR_DBS_PREFIX}\xff"

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

@ -8,21 +8,22 @@ 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 MilvusVectorIOConfig(BaseModel):
db_path: str
kvstore: KVStoreConfig = Field(description="Config for KV store backend (SQLite only for now)")
persistence: KVStoreReference = Field(description="Config for KV store backend (SQLite only for now)")
consistency_level: str = Field(description="The consistency level of the Milvus server", default="Strong")
@classmethod
def sample_run_config(cls, __distro_dir__: str, **kwargs: Any) -> dict[str, Any]:
return {
"db_path": "${env.MILVUS_DB_PATH:=" + __distro_dir__ + "}/" + "milvus.db",
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="milvus_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::milvus",
).model_dump(exclude_none=True),
}

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

@ -9,20 +9,21 @@ 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
@json_schema_type
class QdrantVectorIOConfig(BaseModel):
path: str
kvstore: KVStoreConfig
persistence: KVStoreReference
@classmethod
def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:
return {
"path": "${env.QDRANT_PATH:=~/.llama/" + __distro_dir__ + "}/" + "qdrant.db",
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="qdrant_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::qdrant",
).model_dump(exclude_none=True),
}

11
llama_stack/providers/inline/vector_io/sqlite_vec/config.py
View file

@ -8,18 +8,19 @@ 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
class SQLiteVectorIOConfig(BaseModel):
db_path: str = Field(description="Path to the SQLite database file")
kvstore: KVStoreConfig = Field(description="Config for KV store backend (SQLite only for now)")
persistence: KVStoreReference = Field(description="Config for KV store backend (SQLite only for now)")
@classmethod
def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:
return {
"db_path": "${env.SQLITE_STORE_DIR:=" + __distro_dir__ + "}/" + "sqlite_vec.db",
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__, db_name="sqlite_vec_registry.db"
),
"persistence": KVStoreReference(
backend="kv_default",
namespace="vector_io::sqlite_vec",
).model_dump(exclude_none=True),
}

2
llama_stack/providers/inline/vector_io/sqlite_vec/sqlite_vec.py
View file

@ -389,7 +389,7 @@ class SQLiteVecVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtoc
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)
start_key = VECTOR_DBS_PREFIX
end_key = f"{VECTOR_DBS_PREFIX}\xff"