register model wrapper using alias

src/agentscope/models/__init__.py

@@ -34,14 +34,6 @@
 
 _MODEL_CONFIGS: dict[str, dict] = {}
 
-_MODEL_MAP: dict[str, Type[ModelWrapperBase]] = {
-    "openai": OpenAIChatWrapper,
-    "openai_dall_e": OpenAIDALLEWrapper,
-    "openai_embedding": OpenAIEmbeddingWrapper,
-    "post_api": PostAPIModelWrapperBase,
-    "post_api_chat": PostAPIChatWrapper,
-}
-
 
 def _get_model_wrapper(model_type: str) -> Type[ModelWrapperBase]:
     """Get the specific type of model wrapper
@@ -52,8 +44,10 @@ def _get_model_wrapper(model_type: str) -> Type[ModelWrapperBase]:
     Returns:
         `Type[ModelWrapperBase]`: The corresponding model wrapper class.
     """
-    if model_type in _MODEL_MAP:
-        return _MODEL_MAP[model_type]
+    if model_type in ModelWrapperBase.alias_registry:
+        return ModelWrapperBase.alias_registry[  # type: ignore [return-value]
+            model_type
+        ]
     elif model_type in ModelWrapperBase.registry:
         return ModelWrapperBase.registry[  # type: ignore [return-value]
             model_type

src/agentscope/models/model.py

@@ -190,8 +190,11 @@ def __new__(mcs, name: Any, bases: Any, attrs: Any) -> Any:
     def __init__(cls, name: Any, bases: Any, attrs: Any) -> None:
         if not hasattr(cls, "registry"):
             cls.registry = {}
+            cls.alias_registry = {}
         else:
             cls.registry[name] = cls
+            if hasattr(cls, "alias"):
+                cls.alias_registry[cls.alias] = cls
         super().__init__(name, bases, attrs)
 
 

src/agentscope/models/openai_model.py

@@ -126,6 +126,8 @@ def _metric(self, metric_name: str) -> str:
 class OpenAIChatWrapper(OpenAIWrapper):
     """The model wrapper for OpenAI's chat API."""
 
+    alias: str = "openai"
+
     def _register_default_metrics(self) -> None:
         # Set monitor accordingly
         # TODO: set quota to the following metrics
@@ -234,6 +236,8 @@ def __call__(
 class OpenAIDALLEWrapper(OpenAIWrapper):
     """The model wrapper for OpenAI's DALL·E API."""
 
+    alias: str = "openai_dall_e"
+
     _resolutions: list = [
         "1792*1024",
         "1024*1792",
@@ -330,6 +334,8 @@ def __call__(
 class OpenAIEmbeddingWrapper(OpenAIWrapper):
     """The model wrapper for OpenAI embedding API."""
 
+    alias: str = "openai_embedding"
+
     def _register_default_metrics(self) -> None:
         # Set monitor accordingly
         # TODO: set quota to the following metrics

src/agentscope/models/post_model.py

@@ -16,6 +16,8 @@
 class PostAPIModelWrapperBase(ModelWrapperBase):
     """The base model wrapper for the model deployed on the POST API."""
 
+    alias: str = "post_api"
+
     def __init__(
         self,
         model_id: str,
@@ -172,6 +174,8 @@ class PostAPIChatWrapper(PostAPIModelWrapperBase):
     """A post api model wrapper compatilble with openai chat, e.g., vLLM,
     FastChat."""
 
+    alias: str = "post_api_chat"
+
     def _parse_response(self, response: dict) -> ModelResponse:
         return ModelResponse(
             text=response["data"]["response"]["choices"][0]["message"][
@@ -183,6 +187,8 @@ def _parse_response(self, response: dict) -> ModelResponse:
 class PostAPIDALLEWrapper(PostAPIModelWrapperBase):
     """A post api model wrapper compatible with openai dalle"""
 
+    alias: str = "post_api_dalle"
+
     def _parse_response(self, response: dict) -> ModelResponse:
         urls = [img["url"] for img in response["data"]["response"]["data"]]
         return ModelResponse(image_urls=urls)