phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-19 22:39:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'main' into clarifai-inference-provider

Browse source
This commit is contained in:
sanjaychelliah 2024-11-26 18:01:45 +05:30 • committed by GitHub
commit 4b9085d312
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
536 changed files with 34661 additions and 12116 deletions
Download patch file Download diff file Expand all files Collapse all files

11
llama_stack/providers/registry/agents.py
View file

@ -14,7 +14,7 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.agents,
provider_type="meta-reference",
provider_type="inline::meta-reference",
pip_packages=[
"matplotlib",
"pillow",
@ -22,12 +22,13 @@ def available_providers() -> List[ProviderSpec]:
"scikit-learn",
]
+ kvstore_dependencies(),
module="llama_stack.providers.impls.meta_reference.agents",
config_class="llama_stack.providers.impls.meta_reference.agents.MetaReferenceAgentsImplConfig",
module="llama_stack.providers.inline.agents.meta_reference",
config_class="llama_stack.providers.inline.agents.meta_reference.MetaReferenceAgentsImplConfig",
api_dependencies=[
Api.inference,
Api.safety,
Api.memory,
Api.memory_banks,
],
),
remote_provider_spec(
@ -35,8 +36,8 @@ def available_providers() -> List[ProviderSpec]:
adapter=AdapterSpec(
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.agents.sample",
config_class="llama_stack.providers.adapters.agents.sample.SampleConfig",
module="llama_stack.providers.remote.agents.sample",
config_class="llama_stack.providers.remote.agents.sample.SampleConfig",
),
),
]

33
llama_stack/providers/registry/datasetio.py Normal file
View file

@ -0,0 +1,33 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import List
from llama_stack.distribution.datatypes import * # noqa: F403
def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.datasetio,
provider_type="inline::localfs",
pip_packages=["pandas"],
module="llama_stack.providers.inline.datasetio.localfs",
config_class="llama_stack.providers.inline.datasetio.localfs.LocalFSDatasetIOConfig",
api_dependencies=[],
),
remote_provider_spec(
api=Api.datasetio,
adapter=AdapterSpec(
adapter_type="huggingface",
pip_packages=[
"datasets",
],
module="llama_stack.providers.remote.datasetio.huggingface",
config_class="llama_stack.providers.remote.datasetio.huggingface.HuggingfaceDatasetIOConfig",
),
),
]

28
llama_stack/providers/registry/eval.py Normal file
View file

@ -0,0 +1,28 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import List
from llama_stack.distribution.datatypes import * # noqa: F403
def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.eval,
provider_type="inline::meta-reference",
pip_packages=[],
module="llama_stack.providers.inline.eval.meta_reference",
config_class="llama_stack.providers.inline.eval.meta_reference.MetaReferenceEvalConfig",
api_dependencies=[
Api.datasetio,
Api.datasets,
Api.scoring,
Api.inference,
Api.agents,
],
),
]

113
llama_stack/providers/registry/inference.py
View file

@ -9,40 +9,74 @@ from typing import List
from llama_stack.distribution.datatypes import * # noqa: F403
META_REFERENCE_DEPS = [
"accelerate",
"blobfile",
"fairscale",
"torch",
"torchvision",
"transformers",
"zmq",
"lm-format-enforcer",
]
def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.inference,
provider_type="meta-reference",
provider_type="inline::meta-reference",
pip_packages=META_REFERENCE_DEPS,
module="llama_stack.providers.inline.inference.meta_reference",
config_class="llama_stack.providers.inline.inference.meta_reference.MetaReferenceInferenceConfig",
),
InlineProviderSpec(
api=Api.inference,
provider_type="inline::meta-reference-quantized",
pip_packages=(
META_REFERENCE_DEPS
+ [
"fbgemm-gpu",
"torchao==0.5.0",
]
),
module="llama_stack.providers.inline.inference.meta_reference",
config_class="llama_stack.providers.inline.inference.meta_reference.MetaReferenceQuantizedInferenceConfig",
),
InlineProviderSpec(
api=Api.inference,
provider_type="inline::vllm",
pip_packages=[
"accelerate",
"blobfile",
"fairscale",
"fbgemm-gpu==0.8.0",
"torch",
"torchvision",
"transformers",
"zmq",
"vllm",
],
module="llama_stack.providers.impls.meta_reference.inference",
config_class="llama_stack.providers.impls.meta_reference.inference.MetaReferenceImplConfig",
module="llama_stack.providers.inline.inference.vllm",
config_class="llama_stack.providers.inline.inference.vllm.VLLMConfig",
),
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.inference.sample",
config_class="llama_stack.providers.adapters.inference.sample.SampleConfig",
module="llama_stack.providers.remote.inference.sample",
config_class="llama_stack.providers.remote.inference.sample.SampleConfig",
),
),
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_type="ollama",
pip_packages=["ollama"],
config_class="llama_stack.providers.adapters.inference.ollama.OllamaImplConfig",
module="llama_stack.providers.adapters.inference.ollama",
pip_packages=["ollama", "aiohttp"],
config_class="llama_stack.providers.remote.inference.ollama.OllamaImplConfig",
module="llama_stack.providers.remote.inference.ollama",
),
),
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_type="vllm",
pip_packages=["openai"],
module="llama_stack.providers.remote.inference.vllm",
config_class="llama_stack.providers.remote.inference.vllm.VLLMInferenceAdapterConfig",
),
),
remote_provider_spec(
@ -50,8 +84,8 @@ def available_providers() -> List[ProviderSpec]:
adapter=AdapterSpec(
adapter_type="tgi",
pip_packages=["huggingface_hub", "aiohttp"],
module="llama_stack.providers.adapters.inference.tgi",
config_class="llama_stack.providers.adapters.inference.tgi.TGIImplConfig",
module="llama_stack.providers.remote.inference.tgi",
config_class="llama_stack.providers.remote.inference.tgi.TGIImplConfig",
),
),
remote_provider_spec(
@ -59,8 +93,8 @@ def available_providers() -> List[ProviderSpec]:
adapter=AdapterSpec(
adapter_type="hf::serverless",
pip_packages=["huggingface_hub", "aiohttp"],
module="llama_stack.providers.adapters.inference.tgi",
config_class="llama_stack.providers.adapters.inference.tgi.InferenceAPIImplConfig",
module="llama_stack.providers.remote.inference.tgi",
config_class="llama_stack.providers.remote.inference.tgi.InferenceAPIImplConfig",
),
),
remote_provider_spec(
@ -68,8 +102,8 @@ def available_providers() -> List[ProviderSpec]:
adapter=AdapterSpec(
adapter_type="hf::endpoint",
pip_packages=["huggingface_hub", "aiohttp"],
module="llama_stack.providers.adapters.inference.tgi",
config_class="llama_stack.providers.adapters.inference.tgi.InferenceEndpointImplConfig",
module="llama_stack.providers.remote.inference.tgi",
config_class="llama_stack.providers.remote.inference.tgi.InferenceEndpointImplConfig",
),
),
remote_provider_spec(
@ -79,8 +113,9 @@ def available_providers() -> List[ProviderSpec]:
pip_packages=[
"fireworks-ai",
],
module="llama_stack.providers.adapters.inference.fireworks",
config_class="llama_stack.providers.adapters.inference.fireworks.FireworksImplConfig",
module="llama_stack.providers.remote.inference.fireworks",
config_class="llama_stack.providers.remote.inference.fireworks.FireworksImplConfig",
provider_data_validator="llama_stack.providers.remote.inference.fireworks.FireworksProviderDataValidator",
),
),
remote_provider_spec(
@ -90,9 +125,9 @@ def available_providers() -> List[ProviderSpec]:
pip_packages=[
"together",
],
module="llama_stack.providers.adapters.inference.together",
config_class="llama_stack.providers.adapters.inference.together.TogetherImplConfig",
provider_data_validator="llama_stack.providers.adapters.safety.together.TogetherProviderDataValidator",
module="llama_stack.providers.remote.inference.together",
config_class="llama_stack.providers.remote.inference.together.TogetherImplConfig",
provider_data_validator="llama_stack.providers.remote.inference.together.TogetherProviderDataValidator",
),
),
remote_provider_spec(
@ -100,8 +135,8 @@ def available_providers() -> List[ProviderSpec]:
adapter=AdapterSpec(
adapter_type="bedrock",
pip_packages=["boto3"],
module="llama_stack.providers.adapters.inference.bedrock",
config_class="llama_stack.providers.adapters.inference.bedrock.BedrockConfig",
module="llama_stack.providers.remote.inference.bedrock",
config_class="llama_stack.providers.remote.inference.bedrock.BedrockConfig",
),
),
remote_provider_spec(
@ -111,8 +146,8 @@ def available_providers() -> List[ProviderSpec]:
pip_packages=[
"openai",
],
module="llama_stack.providers.adapters.inference.databricks",
config_class="llama_stack.providers.adapters.inference.databricks.DatabricksImplConfig",
module="llama_stack.providers.remote.inference.databricks",
config_class="llama_stack.providers.remote.inference.databricks.DatabricksImplConfig",
),
),
remote_provider_spec(
@ -126,13 +161,15 @@ def available_providers() -> List[ProviderSpec]:
config_class="llama_stack.providers.adapters.inference.clarifai.ClarifaiImplConfig",
),
),
InlineProviderSpec(
remote_provider_spec(
api=Api.inference,
provider_type="vllm",
pip_packages=[
"vllm",
],
module="llama_stack.providers.impls.vllm",
config_class="llama_stack.providers.impls.vllm.VLLMConfig",
adapter=AdapterSpec(
adapter_type="nvidia",
pip_packages=[
"openai",
],
module="llama_stack.providers.remote.inference.nvidia",
config_class="llama_stack.providers.remote.inference.nvidia.NVIDIAConfig",
),
),
]

39
llama_stack/providers/registry/memory.py
View file

@ -34,17 +34,26 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.memory,
provider_type="meta-reference",
provider_type="inline::meta-reference",
pip_packages=EMBEDDING_DEPS + ["faiss-cpu"],
module="llama_stack.providers.impls.meta_reference.memory",
config_class="llama_stack.providers.impls.meta_reference.memory.FaissImplConfig",
module="llama_stack.providers.inline.memory.faiss",
config_class="llama_stack.providers.inline.memory.faiss.FaissImplConfig",
deprecation_warning="Please use the `inline::faiss` provider instead.",
),
InlineProviderSpec(
api=Api.memory,
provider_type="inline::faiss",
pip_packages=EMBEDDING_DEPS + ["faiss-cpu"],
module="llama_stack.providers.inline.memory.faiss",
config_class="llama_stack.providers.inline.memory.faiss.FaissImplConfig",
),
remote_provider_spec(
Api.memory,
AdapterSpec(
adapter_type="chromadb",
pip_packages=EMBEDDING_DEPS + ["chromadb-client"],
module="llama_stack.providers.adapters.memory.chroma",
module="llama_stack.providers.remote.memory.chroma",
config_class="llama_stack.distribution.datatypes.RemoteProviderConfig",
),
),
remote_provider_spec(
@ -52,8 +61,8 @@ def available_providers() -> List[ProviderSpec]:
AdapterSpec(
adapter_type="pgvector",
pip_packages=EMBEDDING_DEPS + ["psycopg2-binary"],
module="llama_stack.providers.adapters.memory.pgvector",
config_class="llama_stack.providers.adapters.memory.pgvector.PGVectorConfig",
module="llama_stack.providers.remote.memory.pgvector",
config_class="llama_stack.providers.remote.memory.pgvector.PGVectorConfig",
),
),
remote_provider_spec(
@ -61,8 +70,9 @@ def available_providers() -> List[ProviderSpec]:
AdapterSpec(
adapter_type="weaviate",
pip_packages=EMBEDDING_DEPS + ["weaviate-client"],
module="llama_stack.providers.adapters.memory.weaviate",
provider_data_validator="llama_stack.providers.adapters.memory.weaviate.WeaviateRequestProviderData",
module="llama_stack.providers.remote.memory.weaviate",
config_class="llama_stack.providers.remote.memory.weaviate.WeaviateConfig",
provider_data_validator="llama_stack.providers.remote.memory.weaviate.WeaviateRequestProviderData",
),
),
remote_provider_spec(
@ -70,8 +80,17 @@ def available_providers() -> List[ProviderSpec]:
adapter=AdapterSpec(
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.memory.sample",
config_class="llama_stack.providers.adapters.memory.sample.SampleConfig",
module="llama_stack.providers.remote.memory.sample",
config_class="llama_stack.providers.remote.memory.sample.SampleConfig",
),
),
remote_provider_spec(
Api.memory,
AdapterSpec(
adapter_type="qdrant",
pip_packages=EMBEDDING_DEPS + ["qdrant-client"],
module="llama_stack.providers.remote.memory.qdrant",
config_class="llama_stack.providers.remote.memory.qdrant.QdrantConfig",
),
),
]

64
llama_stack/providers/registry/safety.py
View file

@ -19,25 +19,61 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.safety,
provider_type="meta-reference",
provider_type="inline::prompt-guard",
pip_packages=[
"codeshield",
"transformers",
"torch --index-url https://download.pytorch.org/whl/cpu",
],
module="llama_stack.providers.impls.meta_reference.safety",
config_class="llama_stack.providers.impls.meta_reference.safety.SafetyConfig",
module="llama_stack.providers.inline.safety.prompt_guard",
config_class="llama_stack.providers.inline.safety.prompt_guard.PromptGuardConfig",
),
InlineProviderSpec(
api=Api.safety,
provider_type="inline::meta-reference",
pip_packages=[
"transformers",
"torch --index-url https://download.pytorch.org/whl/cpu",
],
module="llama_stack.providers.inline.safety.meta_reference",
config_class="llama_stack.providers.inline.safety.meta_reference.SafetyConfig",
api_dependencies=[
Api.inference,
],
deprecation_error="""
Provider `inline::meta-reference` for API `safety` does not work with the latest Llama Stack.
- if you are using Llama Guard v3, please use the `inline::llama-guard` provider instead.
- if you are using Prompt Guard, please use the `inline::prompt-guard` provider instead.
- if you are using Code Scanner, please use the `inline::code-scanner` provider instead.
""",
),
InlineProviderSpec(
api=Api.safety,
provider_type="inline::llama-guard",
pip_packages=[],
module="llama_stack.providers.inline.safety.llama_guard",
config_class="llama_stack.providers.inline.safety.llama_guard.LlamaGuardConfig",
api_dependencies=[
Api.inference,
],
),
InlineProviderSpec(
api=Api.safety,
provider_type="inline::code-scanner",
pip_packages=[
"codeshield",
],
module="llama_stack.providers.inline.safety.code_scanner",
config_class="llama_stack.providers.inline.safety.code_scanner.CodeScannerConfig",
),
remote_provider_spec(
api=Api.safety,
adapter=AdapterSpec(
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.safety.sample",
config_class="llama_stack.providers.adapters.safety.sample.SampleConfig",
module="llama_stack.providers.remote.safety.sample",
config_class="llama_stack.providers.remote.safety.sample.SampleConfig",
),
),
remote_provider_spec(
@ -45,20 +81,8 @@ def available_providers() -> List[ProviderSpec]:
adapter=AdapterSpec(
adapter_type="bedrock",
pip_packages=["boto3"],
module="llama_stack.providers.adapters.safety.bedrock",
config_class="llama_stack.providers.adapters.safety.bedrock.BedrockSafetyConfig",
),
),
remote_provider_spec(
api=Api.safety,
adapter=AdapterSpec(
adapter_type="together",
pip_packages=[
"together",
],
module="llama_stack.providers.adapters.safety.together",
config_class="llama_stack.providers.adapters.safety.together.TogetherSafetyConfig",
provider_data_validator="llama_stack.providers.adapters.safety.together.TogetherProviderDataValidator",
module="llama_stack.providers.remote.safety.bedrock",
config_class="llama_stack.providers.remote.safety.bedrock.BedrockSafetyConfig",
),
),
]

48
llama_stack/providers/registry/scoring.py Normal file
View file

@ -0,0 +1,48 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import List
from llama_stack.distribution.datatypes import * # noqa: F403
def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.scoring,
provider_type="inline::basic",
pip_packages=[],
module="llama_stack.providers.inline.scoring.basic",
config_class="llama_stack.providers.inline.scoring.basic.BasicScoringConfig",
api_dependencies=[
Api.datasetio,
Api.datasets,
],
),
InlineProviderSpec(
api=Api.scoring,
provider_type="inline::llm-as-judge",
pip_packages=[],
module="llama_stack.providers.inline.scoring.llm_as_judge",
config_class="llama_stack.providers.inline.scoring.llm_as_judge.LlmAsJudgeScoringConfig",
api_dependencies=[
Api.datasetio,
Api.datasets,
Api.inference,
],
),
InlineProviderSpec(
api=Api.scoring,
provider_type="inline::braintrust",
pip_packages=["autoevals", "openai"],
module="llama_stack.providers.inline.scoring.braintrust",
config_class="llama_stack.providers.inline.scoring.braintrust.BraintrustScoringConfig",
api_dependencies=[
Api.datasetio,
Api.datasets,
],
),
]

14
llama_stack/providers/registry/telemetry.py
View file

@ -13,18 +13,18 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.telemetry,
provider_type="meta-reference",
provider_type="inline::meta-reference",
pip_packages=[],
module="llama_stack.providers.impls.meta_reference.telemetry",
config_class="llama_stack.providers.impls.meta_reference.telemetry.ConsoleConfig",
module="llama_stack.providers.inline.meta_reference.telemetry",
config_class="llama_stack.providers.inline.meta_reference.telemetry.ConsoleConfig",
),
remote_provider_spec(
api=Api.telemetry,
adapter=AdapterSpec(
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.telemetry.sample",
config_class="llama_stack.providers.adapters.telemetry.sample.SampleConfig",
module="llama_stack.providers.remote.telemetry.sample",
config_class="llama_stack.providers.remote.telemetry.sample.SampleConfig",
),
),
remote_provider_spec(
@ -37,8 +37,8 @@ def available_providers() -> List[ProviderSpec]:
"opentelemetry-exporter-jaeger",
"opentelemetry-semantic-conventions",
],
module="llama_stack.providers.adapters.telemetry.opentelemetry",
config_class="llama_stack.providers.adapters.telemetry.opentelemetry.OpenTelemetryConfig",
module="llama_stack.providers.remote.telemetry.opentelemetry",
config_class="llama_stack.providers.remote.telemetry.opentelemetry.OpenTelemetryConfig",
),
),
]