Refactor LlamaCpp initialization to simplify parameter handling and remove unused callback manager

src/voice_dialogue/config/llm_config.py

@@ -19,6 +19,8 @@ def get_llm_model_params() -> Dict[str, Any]:
     # 基础模型参数
     model_params = {
         'streaming': True,
+        'n_gpu_layers': -1,
+        'n_batch': 1024,
         'temperature': 0.7,
         'top_p': 0.9,
         'top_k': 20,

src/voice_dialogue/services/text/processor.py

@@ -2,7 +2,6 @@ import pathlib
 import typing
 
 from langchain_community.chat_models.llamacpp import ChatLlamaCpp
-from langchain_core.callbacks import StreamingStdOutCallbackHandler, CallbackManager
 from langchain_core.messages import SystemMessage
 from langchain_core.prompts import (
     ChatPromptTemplate, MessagesPlaceholder, HumanMessagePromptTemplate
@@ -22,20 +21,9 @@ def create_langchain_chat_llamacpp_instance(
     logger.info(">>>>>>> Initializing LlamaCpp Langchain instance...")
 
     model_path = pathlib.Path(local_model_path)
-    callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
     llamacpp_langchain_instance = ChatLlamaCpp(
         model_path=str(model_path),
-        streaming=model_params.get('streaming', True),
-        n_gpu_layers=model_params.get('n_gpu_layers', -1),
-        n_batch=model_params.get('n_batch', 512),
-        n_ctx=model_params.get('n_ctx', 2048),
-        f16_kv=model_params.get('f16_kv', True),
-        temperature=model_params.get('temperature', 0.8),
-        top_k=model_params.get('top_k', 40),
-        top_p=model_params.get('top_p', 0.95),
-        max_tokens=model_params.get('n_predict', 256),
-        # callback_manager=callback_manager,
-        verbose=False
+        **model_params
     )
 
     return llamacpp_langchain_instance