Add provider deprecation support; change directory structure

llama_stack/distribution/resolver.py

@@ -8,6 +8,8 @@ import inspect
 
 from typing import Any, Dict, List, Set
 
+from termcolor import cprint
+
 from llama_stack.providers.datatypes import *  # noqa: F403
 from llama_stack.distribution.datatypes import *  # noqa: F403
 
@@ -97,6 +99,12 @@ async def resolve_impls(
                 )
 
             p = provider_registry[api][provider.provider_type]
+            if p.deprecation_warning:
+                cprint(
+                    f"Provider `{provider.provider_type}` for API `{api}` is deprecated and will be removed in a future release: {p.deprecation_warning}",
+                    "red",
+                    attrs=["bold"],
+                )
             p.deps__ = [a.value for a in p.api_dependencies]
             spec = ProviderWithSpec(
                 spec=p,

llama_stack/providers/datatypes.py

@@ -82,6 +82,10 @@ class ProviderSpec(BaseModel):
         default_factory=list,
         description="Higher-level API surfaces may depend on other providers to provide their functionality",
     )
+    deprecation_warning: Optional[str] = Field(
+        default=None,
+        description="If this provider is deprecated, specify the warning message here",
+    )
 
     # used internally by the resolver; this is a hack for now
     deps__: List[str] = Field(default_factory=list)

llama_stack/providers/inline/agents/meta_reference/config.py

@@ -4,10 +4,9 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from pydantic import BaseModel, Field
-
 from llama_stack.providers.utils.kvstore import KVStoreConfig
 from llama_stack.providers.utils.kvstore.config import SqliteKVStoreConfig
+from pydantic import BaseModel, Field
 
 
 class MetaReferenceAgentsImplConfig(BaseModel):

llama_stack/providers/inline/agents/meta_reference/persistence.py

@@ -11,9 +11,8 @@ from datetime import datetime
 
 from typing import List, Optional
 from llama_stack.apis.agents import *  # noqa: F403
-from pydantic import BaseModel
-
 from llama_stack.providers.utils.kvstore import KVStore
+from pydantic import BaseModel
 
 
 class AgentSessionInfo(BaseModel):

llama_stack/providers/inline/agents/meta_reference/rag/context_retriever.py

@@ -10,14 +10,13 @@ from jinja2 import Template
 from llama_models.llama3.api import *  # noqa: F403
 
 
-from termcolor import cprint  # noqa: F401
-
 from llama_stack.apis.agents import (
     DefaultMemoryQueryGeneratorConfig,
     LLMMemoryQueryGeneratorConfig,
     MemoryQueryGenerator,
     MemoryQueryGeneratorConfig,
 )
+from termcolor import cprint  # noqa: F401
 from llama_stack.apis.inference import *  # noqa: F403
 
 

llama_stack/providers/inline/agents/meta_reference/tools/safety.py

@@ -9,8 +9,7 @@ from typing import List
 from llama_stack.apis.inference import Message
 from llama_stack.apis.safety import *  # noqa: F403
 
-from llama_stack.providers.inline.meta_reference.agents.safety import ShieldRunnerMixin
-
+from ..safety import ShieldRunnerMixin
 from .builtin import BaseTool
 
 

llama_stack/providers/inline/inference/meta_reference/config.py

@@ -10,9 +10,8 @@ from llama_models.datatypes import *  # noqa: F403
 from llama_models.sku_list import resolve_model
 
 from llama_stack.apis.inference import *  # noqa: F401, F403
-from pydantic import BaseModel, Field, field_validator
-
 from llama_stack.providers.utils.inference import supported_inference_models
+from pydantic import BaseModel, Field, field_validator
 
 
 class MetaReferenceInferenceConfig(BaseModel):

llama_stack/providers/inline/inference/meta_reference/generation.py

@@ -35,13 +35,12 @@ from termcolor import cprint
 
 from llama_stack.apis.inference import *  # noqa: F403
 
-from lmformatenforcer import JsonSchemaParser, TokenEnforcer, TokenEnforcerTokenizerData
-
 from llama_stack.distribution.utils.model_utils import model_local_dir
 from llama_stack.providers.utils.inference.prompt_adapter import (
     augment_content_with_response_format_prompt,
     chat_completion_request_to_messages,
 )
+from lmformatenforcer import JsonSchemaParser, TokenEnforcer, TokenEnforcerTokenizerData
 
 from .config import (
     Fp8QuantizationConfig,

llama_stack/providers/inline/inference/meta_reference/parallel_utils.py

@@ -28,13 +28,13 @@ from fairscale.nn.model_parallel.initialize import (
     get_model_parallel_src_rank,
 )
 
+from llama_stack.apis.inference import ChatCompletionRequest, CompletionRequest
+
 from pydantic import BaseModel, Field
 
 from torch.distributed.launcher.api import elastic_launch, LaunchConfig
 from typing_extensions import Annotated
 
-from llama_stack.apis.inference import ChatCompletionRequest, CompletionRequest
-
 from .generation import TokenResult
 
 

llama_stack/providers/inline/inference/meta_reference/quantization/loader.py

@@ -20,16 +20,16 @@ from llama_models.datatypes import CheckpointQuantizationFormat
 from llama_models.llama3.api.args import ModelArgs
 from llama_models.llama3.reference_impl.model import Transformer, TransformerBlock
 from llama_models.sku_list import resolve_model
-from termcolor import cprint
-from torch import nn, Tensor
-
-from torchao.quantization.GPTQ import Int8DynActInt4WeightLinear
 
 from llama_stack.apis.inference import QuantizationType
 
 from llama_stack.providers.inline.meta_reference.inference.config import (
     MetaReferenceQuantizedInferenceConfig,
 )
+from termcolor import cprint
+from torch import nn, Tensor
+
+from torchao.quantization.GPTQ import Int8DynActInt4WeightLinear
 
 
 def swiglu_wrapper(

llama_stack/providers/inline/inference/vllm/config.py

@@ -5,9 +5,9 @@
 # the root directory of this source tree.
 
 from llama_models.schema_utils import json_schema_type
-from pydantic import BaseModel, Field, field_validator
 
 from llama_stack.providers.utils.inference import supported_inference_models
+from pydantic import BaseModel, Field, field_validator
 
 
 @json_schema_type

llama_stack/providers/inline/memory/faiss/config.py

@@ -5,13 +5,13 @@
 # the root directory of this source tree.
 
 from llama_models.schema_utils import json_schema_type
-from pydantic import BaseModel
 
 from llama_stack.distribution.utils.config_dirs import RUNTIME_BASE_DIR
 from llama_stack.providers.utils.kvstore.config import (
     KVStoreConfig,
     SqliteKVStoreConfig,
 )
+from pydantic import BaseModel
 
 
 @json_schema_type

llama_stack/providers/inline/memory/faiss/faiss.py

@@ -8,10 +8,11 @@ import logging
 
 from typing import Any, Dict, List, Optional
 
-import faiss
 import numpy as np
 from numpy.typing import NDArray
 
+import faiss
+
 from llama_models.llama3.api.datatypes import *  # noqa: F403
 
 from llama_stack.apis.memory import *  # noqa: F403

llama_stack/providers/registry/agents.py

@@ -22,8 +22,8 @@ def available_providers() -> List[ProviderSpec]:
                 "scikit-learn",
             ]
             + kvstore_dependencies(),
-            module="llama_stack.providers.inline.meta_reference.agents",
-            config_class="llama_stack.providers.inline.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,

llama_stack/providers/registry/inference.py

@@ -27,8 +27,8 @@ def available_providers() -> List[ProviderSpec]:
             api=Api.inference,
             provider_type="meta-reference",
             pip_packages=META_REFERENCE_DEPS,
-            module="llama_stack.providers.inline.meta_reference.inference",
-            config_class="llama_stack.providers.inline.meta_reference.inference.MetaReferenceInferenceConfig",
+            module="llama_stack.providers.inline.inference.meta_reference",
+            config_class="llama_stack.providers.inline.inference.meta_reference.MetaReferenceInferenceConfig",
         ),
         InlineProviderSpec(
             api=Api.inference,
@@ -40,8 +40,17 @@ def available_providers() -> List[ProviderSpec]:
                     "torchao==0.5.0",
                 ]
             ),
-            module="llama_stack.providers.inline.meta_reference.inference",
-            config_class="llama_stack.providers.inline.meta_reference.inference.MetaReferenceQuantizedInferenceConfig",
+            module="llama_stack.providers.inline.inference.meta_reference",
+            config_class="llama_stack.providers.inline.inference.meta_reference.MetaReferenceQuantizedInferenceConfig",
+        ),
+        InlineProviderSpec(
+            api=Api.inference,
+            provider_type="vllm",
+            pip_packages=[
+                "vllm",
+            ],
+            module="llama_stack.providers.inline.inference.vllm",
+            config_class="llama_stack.providers.inline.inference.vllm.VLLMConfig",
         ),
         remote_provider_spec(
             api=Api.inference,
@@ -140,13 +149,4 @@ def available_providers() -> List[ProviderSpec]:
                 config_class="llama_stack.providers.remote.inference.databricks.DatabricksImplConfig",
             ),
         ),
-        InlineProviderSpec(
-            api=Api.inference,
-            provider_type="vllm",
-            pip_packages=[
-                "vllm",
-            ],
-            module="llama_stack.providers.inline.vllm",
-            config_class="llama_stack.providers.inline.vllm.VLLMConfig",
-        ),
     ]

llama_stack/providers/registry/memory.py

@@ -36,8 +36,16 @@ def available_providers() -> List[ProviderSpec]:
             api=Api.memory,
             provider_type="meta-reference",
             pip_packages=EMBEDDING_DEPS + ["faiss-cpu"],
-            module="llama_stack.providers.inline.meta_reference.memory",
-            config_class="llama_stack.providers.inline.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 `faiss` provider instead.",
+        ),
+        InlineProviderSpec(
+            api=Api.memory,
+            provider_type="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,

llama_stack/providers/tests/agents/fixtures.py

@@ -11,7 +11,7 @@ import pytest_asyncio
 
 from llama_stack.distribution.datatypes import Api, Provider
 
-from llama_stack.providers.inline.meta_reference.agents import (
+from llama_stack.providers.inline.agents.meta_reference import (
     MetaReferenceAgentsImplConfig,
 )
 

llama_stack/providers/tests/inference/fixtures.py

@@ -10,7 +10,7 @@ import pytest
 import pytest_asyncio
 
 from llama_stack.distribution.datatypes import Api, Provider
-from llama_stack.providers.inline.meta_reference.inference import (
+from llama_stack.providers.inline.inference.meta_reference import (
     MetaReferenceInferenceConfig,
 )
 

llama_stack/providers/tests/memory/fixtures.py

@@ -11,7 +11,7 @@ import pytest
 import pytest_asyncio
 
 from llama_stack.distribution.datatypes import Api, Provider
-from llama_stack.providers.inline.meta_reference.memory import FaissImplConfig
+from llama_stack.providers.inline.memory.faiss import FaissImplConfig
 from llama_stack.providers.remote.memory.pgvector import PGVectorConfig
 from llama_stack.providers.remote.memory.weaviate import WeaviateConfig