fix: adapt to agent

services/model_handler.py

@@ -19,20 +19,39 @@ class LocalHuggingFaceModel(Model):
         
     async def ainvoke(self, prompt: str, **kwargs) -> str:
         """Async invoke method"""
-        return self.invoke(prompt=prompt, **kwargs)
+        return await self.invoke(prompt=prompt, **kwargs)
     
     async def ainvoke_stream(self, prompt: str, **kwargs):
         """Async streaming invoke method"""
-        result = self.invoke(prompt=prompt, **kwargs)
+        result = await self.invoke(prompt=prompt, **kwargs)
         yield result
     
     def invoke(self, prompt: str, **kwargs) -> str:
         """Synchronous invoke method"""
-        return self.generate(prompt=prompt, **kwargs)
+        try:
+            inputs = self.tokenizer(prompt, return_tensors="pt", padding=True)
+            
+            #  Configure generation parameters
+            generation_config = {
+                "max_length": self.max_length,
+                "num_return_sequences": 1,
+                "do_sample": kwargs.get("do_sample", False),
+                "temperature": kwargs.get("temperature", 1.0),
+                "top_p": kwargs.get("top_p", 1.0),
+            }
+            
+            # Generate the answer
+            outputs = self.model.generate(**inputs, **generation_config)
+            decoded_output = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
+            
+            return decoded_output
+        except Exception as e:
+            logging.error(f"Error in local model generation: {str(e)}")
+            return f"Error during generation: {str(e)}"
     
     def invoke_stream(self, prompt: str, **kwargs):
         """Synchronous streaming invoke method"""
-        result = self.generate(prompt=prompt, **kwargs)
+        result = self.invoke(prompt=prompt, **kwargs)
         yield result
     
     def parse_provider_response(self, response: str) -> str: