phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-06-27 18:50:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

provider_id => provider_type, adapter_id => adapter_type

Browse source
This commit is contained in:
Ashwin Bharambe 2024-10-02 14:05:59 -07:00
parent df68db644b
commit fe4aabd690
21 changed files with 83 additions and 85 deletions
Download patch file Download diff file Expand all files Collapse all files

12
docs/resources/llama-stack-spec.html
View file

@ -4783,7 +4783,7 @@
"provider_config": {
"type": "object",
"properties": {
"provider_id": {
"provider_type": {
"type": "string"
},
"config": {
@ -4814,7 +4814,7 @@
},
"additionalProperties": false,
"required": [
"provider_id",
"provider_type",
"config"
]
}
@ -4843,7 +4843,7 @@
"provider_config": {
"type": "object",
"properties": {
"provider_id": {
"provider_type": {
"type": "string"
},
"config": {
@ -4874,7 +4874,7 @@
},
"additionalProperties": false,
"required": [
"provider_id",
"provider_type",
"config"
]
}
@ -4894,7 +4894,7 @@
"provider_config": {
"type": "object",
"properties": {
"provider_id": {
"provider_type": {
"type": "string"
},
"config": {
@ -4925,7 +4925,7 @@
},
"additionalProperties": false,
"required": [
"provider_id",
"provider_type",
"config"
]
}

12
docs/resources/llama-stack-spec.yaml
View file

@ -1117,10 +1117,10 @@ components:
- type: array
- type: object
type: object
provider_id:
provider_type:
type: string
required:
- provider_id
- provider_type
- config
type: object
required:
@ -1362,10 +1362,10 @@ components:
- type: array
- type: object
type: object
provider_id:
provider_type:
type: string
required:
- provider_id
- provider_type
- config
type: object
required:
@ -1916,10 +1916,10 @@ components:
- type: array
- type: object
type: object
provider_id:
provider_type:
type: string
required:
- provider_id
- provider_type
- config
type: object
shield_type:

2
llama_stack/apis/memory_banks/memory_banks.py
View file

@ -18,7 +18,7 @@ from llama_stack.distribution.datatypes import GenericProviderConfig
class MemoryBankSpec(BaseModel):
bank_type: MemoryBankType
provider_config: GenericProviderConfig = Field(
description="Provider config for the model, including provider_id, and corresponding config. ",
description="Provider config for the model, including provider_type, and corresponding config. ",
)

2
llama_stack/apis/models/models.py
View file

@ -20,7 +20,7 @@ class ModelServingSpec(BaseModel):
description="All metadatas associated with llama model (defined in llama_models.models.sku_list).",
)
provider_config: GenericProviderConfig = Field(
description="Provider config for the model, including provider_id, and corresponding config. ",
description="Provider config for the model, including provider_type, and corresponding config. ",
)

2
llama_stack/apis/shields/shields.py
View file

@ -16,7 +16,7 @@ from llama_stack.distribution.datatypes import GenericProviderConfig
class ShieldSpec(BaseModel):
shield_type: str
provider_config: GenericProviderConfig = Field(
description="Provider config for the model, including provider_id, and corresponding config. ",
description="Provider config for the model, including provider_type, and corresponding config. ",
)

4
llama_stack/cli/stack/list_providers.py
View file

@ -47,11 +47,11 @@ class StackListProviders(Subcommand):
rows = []
for spec in providers_for_api.values():
if spec.provider_id == "sample":
if spec.provider_type == "sample":
continue
rows.append(
[
spec.provider_id,
spec.provider_type,
",".join(spec.pip_packages),
]
)

10
llama_stack/distribution/configure.py
View file

@ -109,7 +109,7 @@ def configure_api_providers(
routing_entries.append(
RoutableProviderConfig(
routing_key=routing_key,
provider_id=p,
provider_type=p,
config=cfg.dict(),
)
)
@ -120,7 +120,7 @@ def configure_api_providers(
routing_entries.append(
RoutableProviderConfig(
routing_key=[s.value for s in MetaReferenceShieldType],
provider_id=p,
provider_type=p,
config=cfg.dict(),
)
)
@ -133,7 +133,7 @@ def configure_api_providers(
routing_entries.append(
RoutableProviderConfig(
routing_key=routing_key,
provider_id=p,
provider_type=p,
config=cfg.dict(),
)
)
@ -153,7 +153,7 @@ def configure_api_providers(
routing_entries.append(
RoutableProviderConfig(
routing_key=routing_key,
provider_id=p,
provider_type=p,
config=cfg.dict(),
)
)
@ -164,7 +164,7 @@ def configure_api_providers(
)
else:
config.api_providers[api_str] = GenericProviderConfig(
provider_id=p,
provider_type=p,
config=cfg.dict(),
)

2
llama_stack/distribution/datatypes.py
View file

@ -71,7 +71,7 @@ Provider configurations for each of the APIs provided by this package.
E.g. The following is a ProviderRoutingEntry for models:
- routing_key: Meta-Llama3.1-8B-Instruct
provider_id: meta-reference
provider_type: meta-reference
config:
model: Meta-Llama3.1-8B-Instruct
quantization: null

2
llama_stack/distribution/distribution.py
View file

@ -51,7 +51,7 @@ def get_provider_registry() -> Dict[Api, Dict[str, ProviderSpec]]:
module = importlib.import_module(f"llama_stack.providers.registry.{name}")
ret[api] = {
"remote": remote_provider_spec(api),
**{a.provider_id: a for a in module.available_providers()},
**{a.provider_type: a for a in module.available_providers()},
}
return ret

4
llama_stack/distribution/request_headers.py
View file

@ -18,10 +18,10 @@ class NeedsRequestProviderData:
spec = self.__provider_spec__
assert spec, f"Provider spec not set on {self.__class__}"
provider_id = spec.provider_id
provider_type = spec.provider_type
validator_class = spec.provider_data_validator
if not validator_class:
raise ValueError(f"Provider {provider_id} does not have a validator")
raise ValueError(f"Provider {provider_type} does not have a validator")
val = getattr(_THREAD_LOCAL, "provider_data_header_value", None)
if not val:

16
llama_stack/distribution/resolver.py
View file

@ -34,11 +34,11 @@ async def resolve_impls_with_routing(run_config: StackRunConfig) -> Dict[Api, An
if isinstance(config, PlaceholderProviderConfig):
continue
if config.provider_id not in providers:
if config.provider_type not in providers:
raise ValueError(
f"Unknown provider `{config.provider_id}` is not available for API `{api}`"
f"Provider `{config.provider_type}` is not available for API `{api}`"
)
specs[api] = providers[config.provider_id]
specs[api] = providers[config.provider_type]
configs[api] = config
apis_to_serve = run_config.apis_to_serve or set(
@ -68,12 +68,12 @@ async def resolve_impls_with_routing(run_config: StackRunConfig) -> Dict[Api, An
inner_specs = []
inner_deps = []
for rt_entry in routing_table:
if rt_entry.provider_id not in providers:
if rt_entry.provider_type not in providers:
raise ValueError(
f"Unknown provider `{rt_entry.provider_id}` is not available for API `{api}`"
f"Provider `{rt_entry.provider_type}` is not available for API `{api}`"
)
inner_specs.append(providers[rt_entry.provider_id])
inner_deps.extend(providers[rt_entry.provider_id].api_dependencies)
inner_specs.append(providers[rt_entry.provider_type])
inner_deps.extend(providers[rt_entry.provider_type].api_dependencies)
specs[source_api] = RoutingTableProviderSpec(
api=source_api,
@ -94,7 +94,7 @@ async def resolve_impls_with_routing(run_config: StackRunConfig) -> Dict[Api, An
sorted_specs = topological_sort(specs.values())
print(f"Resolved {len(sorted_specs)} providers in topological order")
for spec in sorted_specs:
print(f" {spec.api}: {spec.provider_id}")
print(f" {spec.api}: {spec.provider_type}")
print("")
impls = {}
for spec in sorted_specs:

10
llama_stack/distribution/templates/docker/llamastack-local-cpu/run.yaml
View file

@ -18,7 +18,7 @@ api_providers:
providers:
- meta-reference
agents:
provider_id: meta-reference
provider_type: meta-reference
config:
persistence_store:
namespace: null
@ -28,22 +28,22 @@ api_providers:
providers:
- meta-reference
telemetry:
provider_id: meta-reference
provider_type: meta-reference
config: {}
routing_table:
inference:
- provider_id: remote::ollama
- provider_type: remote::ollama
config:
host: localhost
port: 6000
routing_key: Meta-Llama3.1-8B-Instruct
safety:
- provider_id: meta-reference
- provider_type: meta-reference
config:
llama_guard_shield: null
prompt_guard_shield: null
routing_key: ["llama_guard", "code_scanner_guard", "injection_shield", "jailbreak_shield"]
memory:
- provider_id: meta-reference
- provider_type: meta-reference
config: {}
routing_key: vector

10
llama_stack/distribution/templates/docker/llamastack-local-gpu/run.yaml
View file

@ -18,7 +18,7 @@ api_providers:
providers:
- meta-reference
agents:
provider_id: meta-reference
provider_type: meta-reference
config:
persistence_store:
namespace: null
@ -28,11 +28,11 @@ api_providers:
providers:
- meta-reference
telemetry:
provider_id: meta-reference
provider_type: meta-reference
config: {}
routing_table:
inference:
- provider_id: meta-reference
- provider_type: meta-reference
config:
model: Llama3.1-8B-Instruct
quantization: null
@ -41,12 +41,12 @@ routing_table:
max_batch_size: 1
routing_key: Llama3.1-8B-Instruct
safety:
- provider_id: meta-reference
- provider_type: meta-reference
config:
llama_guard_shield: null
prompt_guard_shield: null
routing_key: ["llama_guard", "code_scanner_guard", "injection_shield", "jailbreak_shield"]
memory:
- provider_id: meta-reference
- provider_type: meta-reference
config: {}
routing_key: vector

4
llama_stack/distribution/utils/dynamic.py
View file

@ -46,11 +46,11 @@ async def instantiate_provider(
assert isinstance(provider_config, List)
routing_table = provider_config
inner_specs = {x.provider_id: x for x in provider_spec.inner_specs}
inner_specs = {x.provider_type: x for x in provider_spec.inner_specs}
inner_impls = []
for routing_entry in routing_table:
impl = await instantiate_provider(
inner_specs[routing_entry.provider_id],
inner_specs[routing_entry.provider_type],
deps,
routing_entry,
)

18
llama_stack/providers/datatypes.py
View file

@ -28,7 +28,7 @@ class Api(Enum):
@json_schema_type
class ProviderSpec(BaseModel):
api: Api
provider_id: str
provider_type: str
config_class: str = Field(
...,
description="Fully-qualified classname of the config for this provider",
@ -56,7 +56,7 @@ class RoutableProvider(Protocol):
class GenericProviderConfig(BaseModel):
provider_id: str
provider_type: str
config: Dict[str, Any]
@ -76,7 +76,7 @@ class RoutableProviderConfig(GenericProviderConfig):
# Example: /inference, /safety
@json_schema_type
class AutoRoutedProviderSpec(ProviderSpec):
provider_id: str = "router"
provider_type: str = "router"
config_class: str = ""
docker_image: Optional[str] = None
@ -101,7 +101,7 @@ class AutoRoutedProviderSpec(ProviderSpec):
# Example: /models, /shields
@json_schema_type
class RoutingTableProviderSpec(ProviderSpec):
provider_id: str = "routing_table"
provider_type: str = "routing_table"
config_class: str = ""
docker_image: Optional[str] = None
@ -119,7 +119,7 @@ class RoutingTableProviderSpec(ProviderSpec):
@json_schema_type
class AdapterSpec(BaseModel):
adapter_id: str = Field(
adapter_type: str = Field(
...,
description="Unique identifier for this adapter",
)
@ -179,8 +179,8 @@ class RemoteProviderConfig(BaseModel):
return f"http://{self.host}:{self.port}"
def remote_provider_id(adapter_id: str) -> str:
return f"remote::{adapter_id}"
def remote_provider_type(adapter_type: str) -> str:
return f"remote::{adapter_type}"
@json_schema_type
@ -226,8 +226,8 @@ def remote_provider_spec(
if adapter and adapter.config_class
else "llama_stack.distribution.datatypes.RemoteProviderConfig"
)
provider_id = remote_provider_id(adapter.adapter_id) if adapter else "remote"
provider_type = remote_provider_type(adapter.adapter_type) if adapter else "remote"
return RemoteProviderSpec(
api=api, provider_id=provider_id, config_class=config_class, adapter=adapter
api=api, provider_type=provider_type, config_class=config_class, adapter=adapter
)

4
llama_stack/providers/registry/agents.py
View file

@ -14,7 +14,7 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.agents,
provider_id="meta-reference",
provider_type="meta-reference",
pip_packages=[
"matplotlib",
"pillow",
@ -33,7 +33,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.agents,
adapter=AdapterSpec(
adapter_id="sample",
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.agents.sample",
config_class="llama_stack.providers.adapters.agents.sample.SampleConfig",

22
llama_stack/providers/registry/inference.py
View file

@ -13,7 +13,7 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.inference,
provider_id="meta-reference",
provider_type="meta-reference",
pip_packages=[
"accelerate",
"blobfile",
@ -30,7 +30,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="sample",
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.inference.sample",
config_class="llama_stack.providers.adapters.inference.sample.SampleConfig",
@ -39,7 +39,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="ollama",
adapter_type="ollama",
pip_packages=["ollama"],
module="llama_stack.providers.adapters.inference.ollama",
),
@ -47,7 +47,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="tgi",
adapter_type="tgi",
pip_packages=["huggingface_hub", "aiohttp"],
module="llama_stack.providers.adapters.inference.tgi",
config_class="llama_stack.providers.adapters.inference.tgi.TGIImplConfig",
@ -56,7 +56,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="hf::serverless",
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",
@ -65,7 +65,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="hf::endpoint",
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",
@ -74,7 +74,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="fireworks",
adapter_type="fireworks",
pip_packages=[
"fireworks-ai",
],
@ -85,7 +85,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="together",
adapter_type="together",
pip_packages=[
"together",
],
@ -97,10 +97,8 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.inference,
adapter=AdapterSpec(
adapter_id="bedrock",
pip_packages=[
"boto3"
],
adapter_type="bedrock",
pip_packages=["boto3"],
module="llama_stack.providers.adapters.inference.bedrock",
config_class="llama_stack.providers.adapters.inference.bedrock.BedrockConfig",
),

8
llama_stack/providers/registry/memory.py
View file

@ -34,7 +34,7 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.memory,
provider_id="meta-reference",
provider_type="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",
@ -42,7 +42,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
Api.memory,
AdapterSpec(
adapter_id="chromadb",
adapter_type="chromadb",
pip_packages=EMBEDDING_DEPS + ["chromadb-client"],
module="llama_stack.providers.adapters.memory.chroma",
),
@ -50,7 +50,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
Api.memory,
AdapterSpec(
adapter_id="pgvector",
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",
@ -59,7 +59,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.memory,
adapter=AdapterSpec(
adapter_id="sample",
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.memory.sample",
config_class="llama_stack.providers.adapters.memory.sample.SampleConfig",

8
llama_stack/providers/registry/safety.py
View file

@ -19,7 +19,7 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.safety,
provider_id="meta-reference",
provider_type="meta-reference",
pip_packages=[
"codeshield",
"transformers",
@ -34,7 +34,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.safety,
adapter=AdapterSpec(
adapter_id="sample",
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.safety.sample",
config_class="llama_stack.providers.adapters.safety.sample.SampleConfig",
@ -43,7 +43,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.safety,
adapter=AdapterSpec(
adapter_id="bedrock",
adapter_type="bedrock",
pip_packages=["boto3"],
module="llama_stack.providers.adapters.safety.bedrock",
config_class="llama_stack.providers.adapters.safety.bedrock.BedrockSafetyConfig",
@ -52,7 +52,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.safety,
adapter=AdapterSpec(
adapter_id="together",
adapter_type="together",
pip_packages=[
"together",
],

6
llama_stack/providers/registry/telemetry.py
View file

@ -13,7 +13,7 @@ def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.telemetry,
provider_id="meta-reference",
provider_type="meta-reference",
pip_packages=[],
module="llama_stack.providers.impls.meta_reference.telemetry",
config_class="llama_stack.providers.impls.meta_reference.telemetry.ConsoleConfig",
@ -21,7 +21,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.telemetry,
adapter=AdapterSpec(
adapter_id="sample",
adapter_type="sample",
pip_packages=[],
module="llama_stack.providers.adapters.telemetry.sample",
config_class="llama_stack.providers.adapters.telemetry.sample.SampleConfig",
@ -30,7 +30,7 @@ def available_providers() -> List[ProviderSpec]:
remote_provider_spec(
api=Api.telemetry,
adapter=AdapterSpec(
adapter_id="opentelemetry-jaeger",
adapter_type="opentelemetry-jaeger",
pip_packages=[
"opentelemetry-api",
"opentelemetry-sdk",

10
tests/examples/local-run.yaml
View file

@ -18,7 +18,7 @@ api_providers:
providers:
- meta-reference
agents:
provider_id: meta-reference
provider_type: meta-reference
config:
persistence_store:
namespace: null
@ -28,11 +28,11 @@ api_providers:
providers:
- meta-reference
telemetry:
provider_id: meta-reference
provider_type: meta-reference
config: {}
routing_table:
inference:
- provider_id: meta-reference
- provider_type: meta-reference
config:
model: Meta-Llama3.1-8B-Instruct
quantization: null
@ -41,7 +41,7 @@ routing_table:
max_batch_size: 1
routing_key: Meta-Llama3.1-8B-Instruct
safety:
- provider_id: meta-reference
- provider_type: meta-reference
config:
llama_guard_shield:
model: Llama-Guard-3-1B
@ -52,6 +52,6 @@ routing_table:
model: Prompt-Guard-86M
routing_key: ["llama_guard", "code_scanner_guard", "injection_shield", "jailbreak_shield"]
memory:
- provider_id: meta-reference
- provider_type: meta-reference
config: {}
routing_key: vector