From 7ccf83fb74136f3d12e4562cc4203576383a0c72 Mon Sep 17 00:00:00 2001 From: Rohan Awhad Date: Wed, 25 Jun 2025 11:10:58 -0400 Subject: [PATCH 1/2] chore: standardize unsupported model error #2517 - llama_stack/exceptions.py: Add UnsupportedModelError class - remote inference ollama.py and utils/inference/model_registry.py: Changed ValueError in favor of UnsupportedModelError - utils/inference/litellm_openai_mixin.py: remote register_model func. Now uses parent class ModelRegistry's func Closes #2517 --- llama_stack/exceptions.py | 13 +++++++++++++ .../providers/remote/inference/ollama/ollama.py | 5 ++--- .../utils/inference/litellm_openai_mixin.py | 7 ------- .../providers/utils/inference/model_registry.py | 5 ++--- 4 files changed, 17 insertions(+), 13 deletions(-) create mode 100644 llama_stack/exceptions.py diff --git a/llama_stack/exceptions.py b/llama_stack/exceptions.py new file mode 100644 index 000000000..80f297bce --- /dev/null +++ b/llama_stack/exceptions.py @@ -0,0 +1,13 @@ +# 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. + + +class UnsupportedModelError(ValueError): + """raised when model is not present in the list of supported models""" + + def __init__(self, model_name: str, supported_models_list: list[str]): + message = f"'{model_name}' model is not supported. Supported models are: {', '.join(supported_models_list)}" + super().__init__(message) diff --git a/llama_stack/providers/remote/inference/ollama/ollama.py b/llama_stack/providers/remote/inference/ollama/ollama.py index d51072fbf..1d66b982f 100644 --- a/llama_stack/providers/remote/inference/ollama/ollama.py +++ b/llama_stack/providers/remote/inference/ollama/ollama.py @@ -51,6 +51,7 @@ from llama_stack.apis.inference.inference import ( OpenAIResponseFormatParam, ) from llama_stack.apis.models import Model, ModelType +from llama_stack.exceptions import UnsupportedModelError from llama_stack.log import get_logger from llama_stack.providers.datatypes import ( HealthResponse, @@ -374,9 +375,7 @@ class OllamaInferenceAdapter( f"Imprecise provider resource id was used but 'latest' is available in Ollama - using '{model.provider_resource_id}:latest'" ) return model - raise ValueError( - f"Model '{model.provider_resource_id}' is not available in Ollama. Available models: {', '.join(available_models)}" - ) + raise UnsupportedModelError(model.provider_resource_id, available_models) model.provider_resource_id = provider_resource_id return model diff --git a/llama_stack/providers/utils/inference/litellm_openai_mixin.py b/llama_stack/providers/utils/inference/litellm_openai_mixin.py index c21f379c9..9fc258880 100644 --- a/llama_stack/providers/utils/inference/litellm_openai_mixin.py +++ b/llama_stack/providers/utils/inference/litellm_openai_mixin.py @@ -40,7 +40,6 @@ from llama_stack.apis.inference.inference import ( OpenAIMessageParam, OpenAIResponseFormatParam, ) -from llama_stack.apis.models.models import Model from llama_stack.distribution.request_headers import NeedsRequestProviderData from llama_stack.log import get_logger from llama_stack.providers.utils.inference.model_registry import ModelRegistryHelper @@ -91,12 +90,6 @@ class LiteLLMOpenAIMixin( async def shutdown(self): pass - async def register_model(self, model: Model) -> Model: - model_id = self.get_provider_model_id(model.provider_resource_id) - if model_id is None: - raise ValueError(f"Unsupported model: {model.provider_resource_id}") - return model - def get_litellm_model_name(self, model_id: str) -> str: # users may be using openai/ prefix in their model names. the openai/models.py did this by default. # model_id.startswith("openai/") is for backwards compatibility. diff --git a/llama_stack/providers/utils/inference/model_registry.py b/llama_stack/providers/utils/inference/model_registry.py index d707e36c2..16ce6ac7c 100644 --- a/llama_stack/providers/utils/inference/model_registry.py +++ b/llama_stack/providers/utils/inference/model_registry.py @@ -9,6 +9,7 @@ from typing import Any from pydantic import BaseModel, Field from llama_stack.apis.models.models import ModelType +from llama_stack.exceptions import UnsupportedModelError from llama_stack.models.llama.sku_list import all_registered_models from llama_stack.providers.datatypes import Model, ModelsProtocolPrivate from llama_stack.providers.utils.inference import ( @@ -81,9 +82,7 @@ class ModelRegistryHelper(ModelsProtocolPrivate): async def register_model(self, model: Model) -> Model: if not (supported_model_id := self.get_provider_model_id(model.provider_resource_id)): - raise ValueError( - f"Model '{model.provider_resource_id}' is not supported. Supported models are: {', '.join(self.alias_to_provider_id_map.keys())}" - ) + raise UnsupportedModelError(model.provider_resource_id, self.alias_to_provider_id_map.keys()) provider_resource_id = self.get_provider_model_id(model.model_id) if model.model_type == ModelType.embedding: # embedding models are always registered by their provider model id and does not need to be mapped to a llama model From 042cb89db221a6b0c7cb4ad04bca4a30c9f1f8b5 Mon Sep 17 00:00:00 2001 From: Rohan Awhad Date: Thu, 26 Jun 2025 10:44:14 -0400 Subject: [PATCH 2/2] fix: reverted to having register_model func and changed error type --- .../providers/utils/inference/litellm_openai_mixin.py | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/llama_stack/providers/utils/inference/litellm_openai_mixin.py b/llama_stack/providers/utils/inference/litellm_openai_mixin.py index 9fc258880..b69cd822f 100644 --- a/llama_stack/providers/utils/inference/litellm_openai_mixin.py +++ b/llama_stack/providers/utils/inference/litellm_openai_mixin.py @@ -40,7 +40,9 @@ from llama_stack.apis.inference.inference import ( OpenAIMessageParam, OpenAIResponseFormatParam, ) +from llama_stack.apis.models.models import Model from llama_stack.distribution.request_headers import NeedsRequestProviderData +from llama_stack.exceptions import UnsupportedModelError from llama_stack.log import get_logger from llama_stack.providers.utils.inference.model_registry import ModelRegistryHelper from llama_stack.providers.utils.inference.openai_compat import ( @@ -90,6 +92,12 @@ class LiteLLMOpenAIMixin( async def shutdown(self): pass + async def register_model(self, model: Model) -> Model: + model_id = self.get_provider_model_id(model.provider_resource_id) + if model_id is None: + raise UnsupportedModelError(model.provider_resource_id, self.alias_to_provider_id_map.keys()) + return model + def get_litellm_model_name(self, model_id: str) -> str: # users may be using openai/ prefix in their model names. the openai/models.py did this by default. # model_id.startswith("openai/") is for backwards compatibility.