phoenix-oss/llama-stack
1
0
Fork 0
forked from phoenix-oss/llama-stack-mirror
Code Pull requests Releases Packages 2 Activity Actions

ModelAlias -> ProviderModelEntry

Browse source
This commit is contained in:
Ashwin Bharambe 2025-02-20 14:02:36 -08:00
parent 561295af76
commit 07ccf908f7
27 changed files with 132 additions and 132 deletions
Download patch file Download diff file Expand all files Collapse all files

2
docs/source/distributions/self_hosted_distro/ollama.md
View file

@ -130,7 +130,7 @@ llama stack run ./run-with-safety.yaml \
### (Optional) Update Model Serving Configuration ### (Optional) Update Model Serving Configuration
```{note} ```{note}
Please check the [model_aliases](https://github.com/meta-llama/llama-stack/blob/main/llama_stack/providers/remote/inference/ollama/ollama.py#L45) for the supported Ollama models. Please check the [model_entries](https://github.com/meta-llama/llama-stack/blob/main/llama_stack/providers/remote/inference/ollama/ollama.py#L45) for the supported Ollama models.
``` ```
To serve a new model with `ollama` To serve a new model with `ollama`

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

@ -46,7 +46,7 @@ from llama_stack.providers.utils.inference.embedding_mixin import (
) )
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
ModelRegistryHelper, ModelRegistryHelper,
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
from llama_stack.providers.utils.inference.prompt_adapter import ( from llama_stack.providers.utils.inference.prompt_adapter import (
augment_content_with_response_format_prompt, augment_content_with_response_format_prompt,
@ -116,7 +116,7 @@ class MetaReferenceInferenceImpl(
self.model_registry_helper = ModelRegistryHelper( self.model_registry_helper = ModelRegistryHelper(
[ [
build_hf_repo_model_alias( build_hf_repo_model_entry(
llama_model.descriptor(), llama_model.descriptor(),
llama_model.core_model_id.value, llama_model.core_model_id.value,
) )

4
llama_stack/providers/remote/inference/bedrock/bedrock.py
View file

@ -43,12 +43,12 @@ from llama_stack.providers.utils.inference.prompt_adapter import (
interleaved_content_as_str, interleaved_content_as_str,
) )
from .models import MODEL_ALIASES from .models import MODEL_ENTRIES
class BedrockInferenceAdapter(ModelRegistryHelper, Inference): class BedrockInferenceAdapter(ModelRegistryHelper, Inference):
def __init__(self, config: BedrockConfig) -> None: def __init__(self, config: BedrockConfig) -> None:
ModelRegistryHelper.__init__(self, MODEL_ALIASES) ModelRegistryHelper.__init__(self, MODEL_ENTRIES)
self._config = config self._config = config
self._client = create_bedrock_client(config) self._client = create_bedrock_client(config)

10
llama_stack/providers/remote/inference/bedrock/models.py
View file

@ -6,19 +6,19 @@
from llama_stack.models.llama.datatypes import CoreModelId from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
MODEL_ALIASES = [ MODEL_ENTRIES = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta.llama3-1-8b-instruct-v1:0", "meta.llama3-1-8b-instruct-v1:0",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta.llama3-1-70b-instruct-v1:0", "meta.llama3-1-70b-instruct-v1:0",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta.llama3-1-405b-instruct-v1:0", "meta.llama3-1-405b-instruct-v1:0",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),

4
llama_stack/providers/remote/inference/cerebras/cerebras.py
View file

@ -41,14 +41,14 @@ from llama_stack.providers.utils.inference.prompt_adapter import (
) )
from .config import CerebrasImplConfig from .config import CerebrasImplConfig
from .models import model_aliases from .models import model_entries
class CerebrasInferenceAdapter(ModelRegistryHelper, Inference): class CerebrasInferenceAdapter(ModelRegistryHelper, Inference):
def __init__(self, config: CerebrasImplConfig) -> None: def __init__(self, config: CerebrasImplConfig) -> None:
ModelRegistryHelper.__init__( ModelRegistryHelper.__init__(
self, self,
model_aliases=model_aliases, model_entries=model_entries,
) )
self.config = config self.config = config

8
llama_stack/providers/remote/inference/cerebras/models.py
View file

@ -6,15 +6,15 @@
from llama_stack.models.llama.datatypes import CoreModelId from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
model_aliases = [ model_entries = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.1-8b", "llama3.1-8b",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama-3.3-70b", "llama-3.3-70b",
CoreModelId.llama3_3_70b_instruct.value, CoreModelId.llama3_3_70b_instruct.value,
), ),

10
llama_stack/providers/remote/inference/databricks/databricks.py
View file

@ -25,7 +25,7 @@ from llama_stack.apis.inference import (
from llama_stack.models.llama.datatypes import CoreModelId from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
ModelRegistryHelper, ModelRegistryHelper,
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
from llama_stack.providers.utils.inference.openai_compat import ( from llama_stack.providers.utils.inference.openai_compat import (
get_sampling_options, get_sampling_options,
@ -38,12 +38,12 @@ from llama_stack.providers.utils.inference.prompt_adapter import (
from .config import DatabricksImplConfig from .config import DatabricksImplConfig
model_aliases = [ model_entries = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"databricks-meta-llama-3-1-70b-instruct", "databricks-meta-llama-3-1-70b-instruct",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"databricks-meta-llama-3-1-405b-instruct", "databricks-meta-llama-3-1-405b-instruct",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),
@ -52,7 +52,7 @@ model_aliases = [
class DatabricksInferenceAdapter(ModelRegistryHelper, Inference): class DatabricksInferenceAdapter(ModelRegistryHelper, Inference):
def __init__(self, config: DatabricksImplConfig) -> None: def __init__(self, config: DatabricksImplConfig) -> None:
ModelRegistryHelper.__init__(self, model_aliases=model_aliases) ModelRegistryHelper.__init__(self, model_entries=model_entries)
self.config = config self.config = config
async def initialize(self) -> None: async def initialize(self) -> None:

4
llama_stack/providers/remote/inference/fireworks/fireworks.py
View file

@ -47,12 +47,12 @@ from llama_stack.providers.utils.inference.prompt_adapter import (
) )
from .config import FireworksImplConfig from .config import FireworksImplConfig
from .models import MODEL_ALIASES from .models import MODEL_ENTRIES
class FireworksInferenceAdapter(ModelRegistryHelper, Inference, NeedsRequestProviderData): class FireworksInferenceAdapter(ModelRegistryHelper, Inference, NeedsRequestProviderData):
def __init__(self, config: FireworksImplConfig) -> None: def __init__(self, config: FireworksImplConfig) -> None:
ModelRegistryHelper.__init__(self, MODEL_ALIASES) ModelRegistryHelper.__init__(self, MODEL_ENTRIES)
self.config = config self.config = config
async def initialize(self) -> None: async def initialize(self) -> None:

24
llama_stack/providers/remote/inference/fireworks/models.py
View file

@ -6,47 +6,47 @@
from llama_stack.models.llama.datatypes import CoreModelId from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
MODEL_ALIASES = [ MODEL_ENTRIES = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p1-8b-instruct", "accounts/fireworks/models/llama-v3p1-8b-instruct",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p1-70b-instruct", "accounts/fireworks/models/llama-v3p1-70b-instruct",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p1-405b-instruct", "accounts/fireworks/models/llama-v3p1-405b-instruct",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p2-1b-instruct", "accounts/fireworks/models/llama-v3p2-1b-instruct",
CoreModelId.llama3_2_1b_instruct.value, CoreModelId.llama3_2_1b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p2-3b-instruct", "accounts/fireworks/models/llama-v3p2-3b-instruct",
CoreModelId.llama3_2_3b_instruct.value, CoreModelId.llama3_2_3b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p2-11b-vision-instruct", "accounts/fireworks/models/llama-v3p2-11b-vision-instruct",
CoreModelId.llama3_2_11b_vision_instruct.value, CoreModelId.llama3_2_11b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p2-90b-vision-instruct", "accounts/fireworks/models/llama-v3p2-90b-vision-instruct",
CoreModelId.llama3_2_90b_vision_instruct.value, CoreModelId.llama3_2_90b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-v3p3-70b-instruct", "accounts/fireworks/models/llama-v3p3-70b-instruct",
CoreModelId.llama3_3_70b_instruct.value, CoreModelId.llama3_3_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-guard-3-8b", "accounts/fireworks/models/llama-guard-3-8b",
CoreModelId.llama_guard_3_8b.value, CoreModelId.llama_guard_3_8b.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"accounts/fireworks/models/llama-guard-3-11b-vision", "accounts/fireworks/models/llama-guard-3-11b-vision",
CoreModelId.llama_guard_3_11b_vision.value, CoreModelId.llama_guard_3_11b_vision.value,
), ),

18
llama_stack/providers/remote/inference/groq/groq.py
View file

@ -31,8 +31,8 @@ from llama_stack.models.llama.sku_list import CoreModelId
from llama_stack.providers.remote.inference.groq.config import GroqConfig from llama_stack.providers.remote.inference.groq.config import GroqConfig
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
ModelRegistryHelper, ModelRegistryHelper,
build_hf_repo_model_alias, build_hf_repo_model_entry,
build_model_alias, build_model_entry,
) )
from .groq_utils import ( from .groq_utils import (
@ -41,20 +41,20 @@ from .groq_utils import (
convert_chat_completion_response_stream, convert_chat_completion_response_stream,
) )
_MODEL_ALIASES = [ _MODEL_ENTRIES = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3-8b-8192", "llama3-8b-8192",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama-3.1-8b-instant", "llama-3.1-8b-instant",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3-70b-8192", "llama3-70b-8192",
CoreModelId.llama3_70b_instruct.value, CoreModelId.llama3_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama-3.3-70b-versatile", "llama-3.3-70b-versatile",
CoreModelId.llama3_3_70b_instruct.value, CoreModelId.llama3_3_70b_instruct.value,
), ),
@ -62,7 +62,7 @@ _MODEL_ALIASES = [
# Preview models aren't recommended for production use, but we include this one # Preview models aren't recommended for production use, but we include this one
# to pass the test fixture # to pass the test fixture
# TODO(aidand): Replace this with a stable model once Groq supports it # TODO(aidand): Replace this with a stable model once Groq supports it
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama-3.2-3b-preview", "llama-3.2-3b-preview",
CoreModelId.llama3_2_3b_instruct.value, CoreModelId.llama3_2_3b_instruct.value,
), ),
@ -73,7 +73,7 @@ class GroqInferenceAdapter(Inference, ModelRegistryHelper, NeedsRequestProviderD
_config: GroqConfig _config: GroqConfig
def __init__(self, config: GroqConfig): def __init__(self, config: GroqConfig):
ModelRegistryHelper.__init__(self, model_aliases=_MODEL_ALIASES) ModelRegistryHelper.__init__(self, model_entries=_MODEL_ENTRIES)
self._config = config self._config = config
def completion( def completion(

22
llama_stack/providers/remote/inference/nvidia/models.py
View file

@ -6,43 +6,43 @@
from llama_stack.models.llama.datatypes import CoreModelId from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
_MODEL_ALIASES = [ _MODEL_ENTRIES = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama3-8b-instruct", "meta/llama3-8b-instruct",
CoreModelId.llama3_8b_instruct.value, CoreModelId.llama3_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama3-70b-instruct", "meta/llama3-70b-instruct",
CoreModelId.llama3_70b_instruct.value, CoreModelId.llama3_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama-3.1-8b-instruct", "meta/llama-3.1-8b-instruct",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama-3.1-70b-instruct", "meta/llama-3.1-70b-instruct",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama-3.1-405b-instruct", "meta/llama-3.1-405b-instruct",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama-3.2-1b-instruct", "meta/llama-3.2-1b-instruct",
CoreModelId.llama3_2_1b_instruct.value, CoreModelId.llama3_2_1b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama-3.2-3b-instruct", "meta/llama-3.2-3b-instruct",
CoreModelId.llama3_2_3b_instruct.value, CoreModelId.llama3_2_3b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama-3.2-11b-vision-instruct", "meta/llama-3.2-11b-vision-instruct",
CoreModelId.llama3_2_11b_vision_instruct.value, CoreModelId.llama3_2_11b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta/llama-3.2-90b-vision-instruct", "meta/llama-3.2-90b-vision-instruct",
CoreModelId.llama3_2_90b_vision_instruct.value, CoreModelId.llama3_2_90b_vision_instruct.value,
), ),

4
llama_stack/providers/remote/inference/nvidia/nvidia.py
View file

@ -33,7 +33,7 @@ from llama_stack.providers.utils.inference.model_registry import (
from llama_stack.providers.utils.inference.prompt_adapter import content_has_media from llama_stack.providers.utils.inference.prompt_adapter import content_has_media
from . import NVIDIAConfig from . import NVIDIAConfig
from .models import _MODEL_ALIASES from .models import _MODEL_ENTRIES
from .openai_utils import ( from .openai_utils import (
convert_chat_completion_request, convert_chat_completion_request,
convert_completion_request, convert_completion_request,
@ -50,7 +50,7 @@ logger = logging.getLogger(__name__)
class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper): class NVIDIAInferenceAdapter(Inference, ModelRegistryHelper):
def __init__(self, config: NVIDIAConfig) -> None: def __init__(self, config: NVIDIAConfig) -> None:
# TODO(mf): filter by available models # TODO(mf): filter by available models
ModelRegistryHelper.__init__(self, model_aliases=_MODEL_ALIASES) ModelRegistryHelper.__init__(self, model_entries=_MODEL_ENTRIES)
logger.info(f"Initializing NVIDIAInferenceAdapter({config.url})...") logger.info(f"Initializing NVIDIAInferenceAdapter({config.url})...")

42
llama_stack/providers/remote/inference/ollama/ollama.py
View file

@ -35,8 +35,8 @@ from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.datatypes import ModelsProtocolPrivate from llama_stack.providers.datatypes import ModelsProtocolPrivate
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
ModelRegistryHelper, ModelRegistryHelper,
build_hf_repo_model_alias, build_hf_repo_model_entry,
build_model_alias, build_model_entry,
) )
from llama_stack.providers.utils.inference.openai_compat import ( from llama_stack.providers.utils.inference.openai_compat import (
OpenAICompatCompletionChoice, OpenAICompatCompletionChoice,
@ -58,74 +58,74 @@ from llama_stack.providers.utils.inference.prompt_adapter import (
log = logging.getLogger(__name__) log = logging.getLogger(__name__)
model_aliases = [ model_entries = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.1:8b-instruct-fp16", "llama3.1:8b-instruct-fp16",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama3.1:8b", "llama3.1:8b",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.1:70b-instruct-fp16", "llama3.1:70b-instruct-fp16",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama3.1:70b", "llama3.1:70b",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.1:405b-instruct-fp16", "llama3.1:405b-instruct-fp16",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama3.1:405b", "llama3.1:405b",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.2:1b-instruct-fp16", "llama3.2:1b-instruct-fp16",
CoreModelId.llama3_2_1b_instruct.value, CoreModelId.llama3_2_1b_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama3.2:1b", "llama3.2:1b",
CoreModelId.llama3_2_1b_instruct.value, CoreModelId.llama3_2_1b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.2:3b-instruct-fp16", "llama3.2:3b-instruct-fp16",
CoreModelId.llama3_2_3b_instruct.value, CoreModelId.llama3_2_3b_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama3.2:3b", "llama3.2:3b",
CoreModelId.llama3_2_3b_instruct.value, CoreModelId.llama3_2_3b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.2-vision:11b-instruct-fp16", "llama3.2-vision:11b-instruct-fp16",
CoreModelId.llama3_2_11b_vision_instruct.value, CoreModelId.llama3_2_11b_vision_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama3.2-vision:latest", "llama3.2-vision:latest",
CoreModelId.llama3_2_11b_vision_instruct.value, CoreModelId.llama3_2_11b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.2-vision:90b-instruct-fp16", "llama3.2-vision:90b-instruct-fp16",
CoreModelId.llama3_2_90b_vision_instruct.value, CoreModelId.llama3_2_90b_vision_instruct.value,
), ),
build_model_alias( build_model_entry(
"llama3.2-vision:90b", "llama3.2-vision:90b",
CoreModelId.llama3_2_90b_vision_instruct.value, CoreModelId.llama3_2_90b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama3.3:70b", "llama3.3:70b",
CoreModelId.llama3_3_70b_instruct.value, CoreModelId.llama3_3_70b_instruct.value,
), ),
# The Llama Guard models don't have their full fp16 versions # The Llama Guard models don't have their full fp16 versions
# so we are going to alias their default version to the canonical SKU # so we are going to alias their default version to the canonical SKU
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama-guard3:8b", "llama-guard3:8b",
CoreModelId.llama_guard_3_8b.value, CoreModelId.llama_guard_3_8b.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"llama-guard3:1b", "llama-guard3:1b",
CoreModelId.llama_guard_3_1b.value, CoreModelId.llama_guard_3_1b.value,
), ),
@ -134,7 +134,7 @@ model_aliases = [
class OllamaInferenceAdapter(Inference, ModelsProtocolPrivate): class OllamaInferenceAdapter(Inference, ModelsProtocolPrivate):
def __init__(self, url: str) -> None: def __init__(self, url: str) -> None:
self.register_helper = ModelRegistryHelper(model_aliases) self.register_helper = ModelRegistryHelper(model_entries)
self.url = url self.url = url
@property @property

22
llama_stack/providers/remote/inference/sambanova/models.py
View file

@ -6,43 +6,43 @@
from llama_stack.models.llama.datatypes import CoreModelId from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
MODEL_ALIASES = [ MODEL_ENTRIES = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Meta-Llama-3.1-8B-Instruct", "Meta-Llama-3.1-8B-Instruct",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Meta-Llama-3.1-70B-Instruct", "Meta-Llama-3.1-70B-Instruct",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Meta-Llama-3.1-405B-Instruct", "Meta-Llama-3.1-405B-Instruct",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Meta-Llama-3.2-1B-Instruct", "Meta-Llama-3.2-1B-Instruct",
CoreModelId.llama3_2_1b_instruct.value, CoreModelId.llama3_2_1b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Meta-Llama-3.2-3B-Instruct", "Meta-Llama-3.2-3B-Instruct",
CoreModelId.llama3_2_3b_instruct.value, CoreModelId.llama3_2_3b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Meta-Llama-3.3-70B-Instruct", "Meta-Llama-3.3-70B-Instruct",
CoreModelId.llama3_3_70b_instruct.value, CoreModelId.llama3_3_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Llama-3.2-11B-Vision-Instruct", "Llama-3.2-11B-Vision-Instruct",
CoreModelId.llama3_2_11b_vision_instruct.value, CoreModelId.llama3_2_11b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Llama-3.2-90B-Vision-Instruct", "Llama-3.2-90B-Vision-Instruct",
CoreModelId.llama3_2_90b_vision_instruct.value, CoreModelId.llama3_2_90b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"Meta-Llama-Guard-3-8B", "Meta-Llama-Guard-3-8B",
CoreModelId.llama_guard_3_8b.value, CoreModelId.llama_guard_3_8b.value,
), ),

4
llama_stack/providers/remote/inference/sambanova/sambanova.py
View file

@ -31,12 +31,12 @@ from llama_stack.providers.utils.inference.prompt_adapter import (
) )
from .config import SambaNovaImplConfig from .config import SambaNovaImplConfig
from .models import MODEL_ALIASES from .models import MODEL_ENTRIES
class SambaNovaInferenceAdapter(ModelRegistryHelper, Inference): class SambaNovaInferenceAdapter(ModelRegistryHelper, Inference):
def __init__(self, config: SambaNovaImplConfig) -> None: def __init__(self, config: SambaNovaImplConfig) -> None:
ModelRegistryHelper.__init__(self, model_aliases=MODEL_ALIASES) ModelRegistryHelper.__init__(self, model_entries=MODEL_ENTRIES)
self.config = config self.config = config
async def initialize(self) -> None: async def initialize(self) -> None:

8
llama_stack/providers/remote/inference/tgi/tgi.py
View file

@ -32,7 +32,7 @@ from llama_stack.models.llama.sku_list import all_registered_models
from llama_stack.providers.datatypes import ModelsProtocolPrivate from llama_stack.providers.datatypes import ModelsProtocolPrivate
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
ModelRegistryHelper, ModelRegistryHelper,
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
from llama_stack.providers.utils.inference.openai_compat import ( from llama_stack.providers.utils.inference.openai_compat import (
OpenAICompatCompletionChoice, OpenAICompatCompletionChoice,
@ -53,9 +53,9 @@ from .config import InferenceAPIImplConfig, InferenceEndpointImplConfig, TGIImpl
log = logging.getLogger(__name__) log = logging.getLogger(__name__)
def build_hf_repo_model_aliases(): def build_hf_repo_model_entries():
return [ return [
build_hf_repo_model_alias( build_hf_repo_model_entry(
model.huggingface_repo, model.huggingface_repo,
model.descriptor(), model.descriptor(),
) )
@ -70,7 +70,7 @@ class _HfAdapter(Inference, ModelsProtocolPrivate):
model_id: str model_id: str
def __init__(self) -> None: def __init__(self) -> None:
self.register_helper = ModelRegistryHelper(build_hf_repo_model_aliases()) self.register_helper = ModelRegistryHelper(build_hf_repo_model_entries())
self.huggingface_repo_to_llama_model_id = { self.huggingface_repo_to_llama_model_id = {
model.huggingface_repo: model.descriptor() for model in all_registered_models() if model.huggingface_repo model.huggingface_repo: model.descriptor() for model in all_registered_models() if model.huggingface_repo
} }

22
llama_stack/providers/remote/inference/together/models.py
View file

@ -6,43 +6,43 @@
from llama_stack.models.llama.datatypes import CoreModelId from llama_stack.models.llama.datatypes import CoreModelId
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
MODEL_ALIASES = [ MODEL_ENTRIES = [
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo", "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo",
CoreModelId.llama3_1_8b_instruct.value, CoreModelId.llama3_1_8b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo", "meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo",
CoreModelId.llama3_1_70b_instruct.value, CoreModelId.llama3_1_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo", "meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo",
CoreModelId.llama3_1_405b_instruct.value, CoreModelId.llama3_1_405b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Llama-3.2-3B-Instruct-Turbo", "meta-llama/Llama-3.2-3B-Instruct-Turbo",
CoreModelId.llama3_2_3b_instruct.value, CoreModelId.llama3_2_3b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Llama-3.2-11B-Vision-Instruct-Turbo", "meta-llama/Llama-3.2-11B-Vision-Instruct-Turbo",
CoreModelId.llama3_2_11b_vision_instruct.value, CoreModelId.llama3_2_11b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo", "meta-llama/Llama-3.2-90B-Vision-Instruct-Turbo",
CoreModelId.llama3_2_90b_vision_instruct.value, CoreModelId.llama3_2_90b_vision_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Llama-3.3-70B-Instruct-Turbo", "meta-llama/Llama-3.3-70B-Instruct-Turbo",
CoreModelId.llama3_3_70b_instruct.value, CoreModelId.llama3_3_70b_instruct.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Meta-Llama-Guard-3-8B", "meta-llama/Meta-Llama-Guard-3-8B",
CoreModelId.llama_guard_3_8b.value, CoreModelId.llama_guard_3_8b.value,
), ),
build_hf_repo_model_alias( build_hf_repo_model_entry(
"meta-llama/Llama-Guard-3-11B-Vision-Turbo", "meta-llama/Llama-Guard-3-11B-Vision-Turbo",
CoreModelId.llama_guard_3_11b_vision.value, CoreModelId.llama_guard_3_11b_vision.value,
), ),

4
llama_stack/providers/remote/inference/together/together.py
View file

@ -46,12 +46,12 @@ from llama_stack.providers.utils.inference.prompt_adapter import (
) )
from .config import TogetherImplConfig from .config import TogetherImplConfig
from .models import MODEL_ALIASES from .models import MODEL_ENTRIES
class TogetherInferenceAdapter(ModelRegistryHelper, Inference, NeedsRequestProviderData): class TogetherInferenceAdapter(ModelRegistryHelper, Inference, NeedsRequestProviderData):
def __init__(self, config: TogetherImplConfig) -> None: def __init__(self, config: TogetherImplConfig) -> None:
ModelRegistryHelper.__init__(self, MODEL_ALIASES) ModelRegistryHelper.__init__(self, MODEL_ENTRIES)
self.config = config self.config = config
async def initialize(self) -> None: async def initialize(self) -> None:

8
llama_stack/providers/remote/inference/vllm/vllm.py
View file

@ -38,7 +38,7 @@ from llama_stack.models.llama.sku_list import all_registered_models
from llama_stack.providers.datatypes import ModelsProtocolPrivate from llama_stack.providers.datatypes import ModelsProtocolPrivate
from llama_stack.providers.utils.inference.model_registry import ( from llama_stack.providers.utils.inference.model_registry import (
ModelRegistryHelper, ModelRegistryHelper,
build_hf_repo_model_alias, build_hf_repo_model_entry,
) )
from llama_stack.providers.utils.inference.openai_compat import ( from llama_stack.providers.utils.inference.openai_compat import (
OpenAICompatCompletionResponse, OpenAICompatCompletionResponse,
@ -62,9 +62,9 @@ from .config import VLLMInferenceAdapterConfig
log = logging.getLogger(__name__) log = logging.getLogger(__name__)
def build_hf_repo_model_aliases(): def build_hf_repo_model_entries():
return [ return [
build_hf_repo_model_alias( build_hf_repo_model_entry(
model.huggingface_repo, model.huggingface_repo,
model.descriptor(), model.descriptor(),
) )
@ -204,7 +204,7 @@ async def _process_vllm_chat_completion_stream_response(
class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate): class VLLMInferenceAdapter(Inference, ModelsProtocolPrivate):
def __init__(self, config: VLLMInferenceAdapterConfig) -> None: def __init__(self, config: VLLMInferenceAdapterConfig) -> None:
self.register_helper = ModelRegistryHelper(build_hf_repo_model_aliases()) self.register_helper = ModelRegistryHelper(build_hf_repo_model_entries())
self.config = config self.config = config
self.client = None self.client = None

14
llama_stack/providers/utils/inference/model_registry.py
View file

@ -18,7 +18,7 @@ from llama_stack.providers.utils.inference import (
# TODO: this class is more confusing than useful right now. We need to make it # TODO: this class is more confusing than useful right now. We need to make it
# more closer to the Model class. # more closer to the Model class.
class ModelAlias(BaseModel): class ProviderModelEntry(BaseModel):
provider_model_id: str provider_model_id: str
aliases: List[str] = Field(default_factory=list) aliases: List[str] = Field(default_factory=list)
llama_model: Optional[str] = None llama_model: Optional[str] = None
@ -32,8 +32,8 @@ def get_huggingface_repo(model_descriptor: str) -> Optional[str]:
return None return None
def build_hf_repo_model_alias(provider_model_id: str, model_descriptor: str) -> ModelAlias: def build_hf_repo_model_entry(provider_model_id: str, model_descriptor: str) -> ProviderModelEntry:
return ModelAlias( return ProviderModelEntry(
provider_model_id=provider_model_id, provider_model_id=provider_model_id,
aliases=[ aliases=[
get_huggingface_repo(model_descriptor), get_huggingface_repo(model_descriptor),
@ -42,8 +42,8 @@ def build_hf_repo_model_alias(provider_model_id: str, model_descriptor: str) ->
) )
def build_model_alias(provider_model_id: str, model_descriptor: str) -> ModelAlias: def build_model_entry(provider_model_id: str, model_descriptor: str) -> ProviderModelEntry:
return ModelAlias( return ProviderModelEntry(
provider_model_id=provider_model_id, provider_model_id=provider_model_id,
aliases=[], aliases=[],
llama_model=model_descriptor, llama_model=model_descriptor,
@ -51,10 +51,10 @@ def build_model_alias(provider_model_id: str, model_descriptor: str) -> ModelAli
class ModelRegistryHelper(ModelsProtocolPrivate): class ModelRegistryHelper(ModelsProtocolPrivate):
def __init__(self, model_aliases: List[ModelAlias]): def __init__(self, model_entries: List[ProviderModelEntry]):
self.alias_to_provider_id_map = {} self.alias_to_provider_id_map = {}
self.provider_id_to_llama_model_map = {} self.provider_id_to_llama_model_map = {}
for alias_obj in model_aliases: for alias_obj in model_entries:
for alias in alias_obj.aliases: for alias in alias_obj.aliases:
self.alias_to_provider_id_map[alias] = alias_obj.provider_model_id self.alias_to_provider_id_map[alias] = alias_obj.provider_model_id
# also add a mapping from provider model id to itself for easy lookup # also add a mapping from provider model id to itself for easy lookup

4
llama_stack/templates/bedrock/bedrock.py
View file

@ -10,7 +10,7 @@ from llama_stack.apis.models import ModelInput
from llama_stack.distribution.datatypes import Provider, ToolGroupInput from llama_stack.distribution.datatypes import Provider, ToolGroupInput
from llama_stack.models.llama.sku_list import all_registered_models from llama_stack.models.llama.sku_list import all_registered_models
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.bedrock.models import MODEL_ALIASES from llama_stack.providers.remote.inference.bedrock.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
@ -47,7 +47,7 @@ def get_distribution_template() -> DistributionTemplate:
provider_model_id=m.provider_model_id, provider_model_id=m.provider_model_id,
provider_id="bedrock", provider_id="bedrock",
) )
for m in MODEL_ALIASES for m in MODEL_ENTRIES
] ]
default_tool_groups = [ default_tool_groups = [
ToolGroupInput( ToolGroupInput(

4
llama_stack/templates/cerebras/cerebras.py
View file

@ -14,7 +14,7 @@ from llama_stack.providers.inline.inference.sentence_transformers import (
) )
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.cerebras import CerebrasImplConfig from llama_stack.providers.remote.inference.cerebras import CerebrasImplConfig
from llama_stack.providers.remote.inference.cerebras.models import model_aliases from llama_stack.providers.remote.inference.cerebras.models import model_entries
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
@ -55,7 +55,7 @@ def get_distribution_template() -> DistributionTemplate:
provider_model_id=m.provider_model_id, provider_model_id=m.provider_model_id,
provider_id="cerebras", provider_id="cerebras",
) )
for m in model_aliases for m in model_entries
] ]
embedding_model = ModelInput( embedding_model = ModelInput(
model_id="all-MiniLM-L6-v2", model_id="all-MiniLM-L6-v2",

4
llama_stack/templates/fireworks/fireworks.py
View file

@ -19,7 +19,7 @@ from llama_stack.providers.inline.inference.sentence_transformers import (
) )
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.fireworks import FireworksImplConfig from llama_stack.providers.remote.inference.fireworks import FireworksImplConfig
from llama_stack.providers.remote.inference.fireworks.models import MODEL_ALIASES from llama_stack.providers.remote.inference.fireworks.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
@ -67,7 +67,7 @@ def get_distribution_template() -> DistributionTemplate:
provider_model_id=m.provider_model_id, provider_model_id=m.provider_model_id,
provider_id="fireworks", provider_id="fireworks",
) )
for m in MODEL_ALIASES for m in MODEL_ENTRIES
] ]
embedding_model = ModelInput( embedding_model = ModelInput(
model_id="all-MiniLM-L6-v2", model_id="all-MiniLM-L6-v2",

4
llama_stack/templates/nvidia/nvidia.py
View file

@ -9,7 +9,7 @@ from pathlib import Path
from llama_stack.distribution.datatypes import ModelInput, Provider, ToolGroupInput from llama_stack.distribution.datatypes import ModelInput, Provider, ToolGroupInput
from llama_stack.models.llama.sku_list import all_registered_models from llama_stack.models.llama.sku_list import all_registered_models
from llama_stack.providers.remote.inference.nvidia import NVIDIAConfig from llama_stack.providers.remote.inference.nvidia import NVIDIAConfig
from llama_stack.providers.remote.inference.nvidia.models import _MODEL_ALIASES from llama_stack.providers.remote.inference.nvidia.models import _MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
@ -45,7 +45,7 @@ def get_distribution_template() -> DistributionTemplate:
provider_model_id=m.provider_model_id, provider_model_id=m.provider_model_id,
provider_id="nvidia", provider_id="nvidia",
) )
for m in _MODEL_ALIASES for m in _MODEL_ENTRIES
] ]
default_tool_groups = [ default_tool_groups = [
ToolGroupInput( ToolGroupInput(

2
llama_stack/templates/ollama/doc_template.md
View file

@ -119,7 +119,7 @@ llama stack run ./run-with-safety.yaml \
### (Optional) Update Model Serving Configuration ### (Optional) Update Model Serving Configuration
```{note} ```{note}
Please check the [model_aliases](https://github.com/meta-llama/llama-stack/blob/main/llama_stack/providers/remote/inference/ollama/ollama.py#L45) for the supported Ollama models. Please check the [model_entries](https://github.com/meta-llama/llama-stack/blob/main/llama_stack/providers/remote/inference/ollama/ollama.py#L45) for the supported Ollama models.
``` ```
To serve a new model with `ollama` To serve a new model with `ollama`

4
llama_stack/templates/sambanova/sambanova.py
View file

@ -14,7 +14,7 @@ from llama_stack.distribution.datatypes import (
) )
from llama_stack.models.llama.sku_list import all_registered_models from llama_stack.models.llama.sku_list import all_registered_models
from llama_stack.providers.remote.inference.sambanova import SambaNovaImplConfig from llama_stack.providers.remote.inference.sambanova import SambaNovaImplConfig
from llama_stack.providers.remote.inference.sambanova.models import MODEL_ALIASES from llama_stack.providers.remote.inference.sambanova.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
@ -47,7 +47,7 @@ def get_distribution_template() -> DistributionTemplate:
provider_model_id=m.provider_model_id, provider_model_id=m.provider_model_id,
provider_id=name, provider_id=name,
) )
for m in MODEL_ALIASES for m in MODEL_ENTRIES
] ]
default_tool_groups = [ default_tool_groups = [

4
llama_stack/templates/together/together.py
View file

@ -19,7 +19,7 @@ from llama_stack.providers.inline.inference.sentence_transformers import (
) )
from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig from llama_stack.providers.inline.vector_io.faiss.config import FaissVectorIOConfig
from llama_stack.providers.remote.inference.together import TogetherImplConfig from llama_stack.providers.remote.inference.together import TogetherImplConfig
from llama_stack.providers.remote.inference.together.models import MODEL_ALIASES from llama_stack.providers.remote.inference.together.models import MODEL_ENTRIES
from llama_stack.templates.template import DistributionTemplate, RunConfigSettings from llama_stack.templates.template import DistributionTemplate, RunConfigSettings
@ -65,7 +65,7 @@ def get_distribution_template() -> DistributionTemplate:
provider_model_id=m.provider_model_id, provider_model_id=m.provider_model_id,
provider_id="together", provider_id="together",
) )
for m in MODEL_ALIASES for m in MODEL_ENTRIES
] ]
default_tool_groups = [ default_tool_groups = [
ToolGroupInput( ToolGroupInput(