feat(model): support specific model parameter in CommonModel

Add specific_model parameter to CommonModel class and implement logic to handle string input for model function. The changes allow specifying a particular model when creating CommonModel instances and ensure the specific model is used when available.

The model function now properly handles string inputs by converting them to appropriate model objects through ModelClient, and includes the specific_model in the returned CommonModel instance.

fixes issue with model selection when using string-based model references

feat(model): 在 CommonModel 中支持 specific model 参数

向 CommonModel 类添加 specific_model 参数并实现处理模型函数的字符串输入的逻辑。这些更改允许在创建 CommonModel 实例时指定特定模型，并确保在可用时使用特定模型。

模型函数现在通过 ModelClient 正确处理字符串输入并将它们转换为适当的模型对象，并在返回的 CommonModel 实例中包含 specific_model。

解决了使用基于字符串的模型引用时的模型选择问题

Change-Id: Ic4c853923510fe28eac7b5b84242e3be23474fa8
Signed-off-by: OhYee <oyohyee@oyohyee.com>

Author: OhYee

agentrun/integration/builtin/model.py

@@ -69,29 +69,27 @@ def model(
     backend_type = kwargs.get("backend_type")
     model = kwargs.get("model")
 
+    if isinstance(input, str):
+        from agentrun.model.client import ModelClient
+
+        client = ModelClient(config=config)
+        input = client.get(name=input, backend_type=backend_type, config=config)
+
     if isinstance(input, ModelProxy):
         return CommonModel(
             model=model or "",
             model_obj=input,
             backend_type=BackendType.PROXY,
+            specific_model=model,
             config=config,
         )
     elif isinstance(input, ModelService):
         return CommonModel(
             model=model or "",
             model_obj=input,
             backend_type=BackendType.SERVICE,
+            specific_model=model,
             config=config,
         )
-
-    from agentrun.model.client import ModelClient
-
-    client = ModelClient(config=config)
-    model_obj = client.get(name=input, backend_type=backend_type, config=config)
-
-    return CommonModel(
-        model=input,
-        model_obj=model_obj,
-        backend_type=backend_type,
-        config=config,
-    )
+    else:
+        raise TypeError("input must be str, ModelProxy or ModelService")

agentrun/integration/utils/model.py

@@ -22,11 +22,13 @@ def __init__(
         model: Optional[str],
         model_obj: Union[ModelService, ModelProxy],
         backend_type: Optional[BackendType] = None,
+        specific_model: Optional[str] = None,
         config: Optional[Config] = None,
     ):
         self.model = model
         self.model_obj = model_obj
         self.backend_type = backend_type
+        self.specific_model = specific_model
         self.config = config or Config()
 
     def completions(self, *args, **kwargs):
@@ -40,7 +42,10 @@ def responses(self, *args, **kwargs):
     def get_model_info(self, config: Optional[Config] = None):
         """获取模型信息"""
         cfg = Config.with_configs(self.config, config)
-        return self.model_obj.model_info(config=cfg)
+        info = self.model_obj.model_info(config=cfg)
+        if self.specific_model:
+            info.model = self.specific_model
+        return info
 
     def __convert_model(self, adapter_name: str):
         try: