better: 优化模型配置和mood

src/chat/heart_flow/heartflow_message_processor.py

@@ -72,16 +72,15 @@ class HeartFCMessageReceiver:
             chat = message.chat_stream
 
             # 2. 兴趣度计算与更新
-            interested_rate, keywords = await _calculate_interest(message)
+            _, keywords = await _calculate_interest(message)
 
             await self.storage.store_message(message, chat)
 
             heartflow_chat: HeartFChatting = await heartflow.get_or_create_heartflow_chat(chat.stream_id)  # type: ignore
 
-            # subheartflow.add_message_to_normal_chat_cache(message, interested_rate, is_mentioned)
             if global_config.mood.enable_mood:
                 chat_mood = mood_manager.get_mood_by_chat_id(heartflow_chat.stream_id)
-                asyncio.create_task(chat_mood.update_mood_by_message(message, interested_rate))
+                asyncio.create_task(chat_mood.update_mood_by_message(message))
 
             # 3. 日志记录
             mes_name = chat.group_info.group_name if chat.group_info else "私聊"
@@ -109,7 +108,7 @@ class HeartFCMessageReceiver:
                 replace_bot_name=True,
             )
 
-            logger.info(f"[{mes_name}]{userinfo.user_nickname}:{processed_plain_text}[{interested_rate:.2f}]")  # type: ignore
+            logger.info(f"[{mes_name}]{userinfo.user_nickname}:{processed_plain_text}")  # type: ignore
 
             _ = Person.register_person(
                 platform=message.message_info.platform,  # type: ignore

src/config/api_ada_configs.py

@@ -102,9 +102,6 @@ class ModelTaskConfig(ConfigBase):
     replyer: TaskConfig
     """normal_chat首要回复模型模型配置"""
 
-    emotion: TaskConfig
-    """情绪模型配置"""
-
     vlm: TaskConfig
     """视觉语言模型配置"""
 

src/mood/mood_manager.py

@@ -62,11 +62,11 @@ class ChatMood:
 
         self.regression_count: int = 0
 
-        self.mood_model = LLMRequest(model_set=model_config.model_task_config.emotion, request_type="mood")
+        self.mood_model = LLMRequest(model_set=model_config.model_task_config.utils, request_type="mood")
 
         self.last_change_time: float = 0
 
-    async def update_mood_by_message(self, message: MessageRecv, interested_rate: float):
+    async def update_mood_by_message(self, message: MessageRecv):
         self.regression_count = 0
 
         during_last_time = message.message_info.time - self.last_change_time  # type: ignore
@@ -74,10 +74,9 @@ class ChatMood:
         base_probability = 0.05
         time_multiplier = 4 * (1 - math.exp(-0.01 * during_last_time))
 
-        if interested_rate <= 0:
-            interest_multiplier = 0
-        else:
-            interest_multiplier = 2 * math.pow(interested_rate, 0.25)
+        # 基于消息长度计算基础兴趣度
+        message_length = len(message.message_content.content or "")
+        interest_multiplier = min(2.0, 1.0 + message_length / 100)
 
         logger.debug(
             f"base_probability: {base_probability}, time_multiplier: {time_multiplier}, interest_multiplier: {interest_multiplier}"
@@ -90,7 +89,7 @@ class ChatMood:
             return
 
         logger.debug(
-            f"{self.log_prefix} 更新情绪状态，感兴趣度: {interested_rate:.2f}, 更新概率: {update_probability:.2f}"
+            f"{self.log_prefix} 更新情绪状态，更新概率: {update_probability:.2f}"
         )
 
         message_time: float = message.message_info.time  # type: ignore

template/model_config_template.toml

@@ -1,5 +1,5 @@
 [inner]
-version = "1.6.0"
+version = "1.7.0"
 
 # 配置文件版本号迭代规则同bot_config.toml
 
@@ -12,14 +12,14 @@ max_retry = 2                           # 最大重试次数（单个模型API
 timeout = 30                            # API请求超时时间（单位：秒）
 retry_interval = 10                     # 重试间隔时间（单位：秒）
 
-[[api_providers]] # SiliconFlow的API服务商配置
-name = "SiliconFlow"
-base_url = "https://api.siliconflow.cn/v1"
-api_key = "your-siliconflow-api-key"
+[[api_providers]] # 阿里 百炼 API服务商配置
+name = "BaiLian"
+base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1"
+api_key = "your-bailian-key"
 client_type = "openai"
 max_retry = 2
-timeout = 30
-retry_interval = 10
+timeout = 15
+retry_interval = 5
 
 [[api_providers]] # 特殊：Google的Gimini使用特殊API，与OpenAI格式不兼容，需要配置client为"gemini"
 name = "Google"
@@ -30,14 +30,14 @@ max_retry = 2
 timeout = 30
 retry_interval = 10
 
-[[api_providers]] # 阿里 百炼 API服务商配置
-name = "BaiLian"
-base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1"
-api_key = "your-bailian-key"
+[[api_providers]] # SiliconFlow的API服务商配置
+name = "SiliconFlow"
+base_url = "https://api.siliconflow.cn/v1"
+api_key = "your-siliconflow-api-key"
 client_type = "openai"
 max_retry = 2
-timeout = 15
-retry_interval = 5
+timeout = 60
+retry_interval = 10
 
 
 [[models]] # 模型（可以配置多个）
@@ -93,8 +93,8 @@ price_in = 0
 price_out = 0
 
 
-[model_task_config.utils] # 在麦麦的一些组件中使用的模型，例如表情包模块，取名模块，关系模块，是麦麦必须的模型
-model_list = ["siliconflow-deepseek-v3"] # 使用的模型列表，每个子项对应上面的模型名称(name)
+[model_task_config.utils] # 在麦麦的一些组件中使用的模型，例如表情包模块，取名模块，关系模块，麦麦的情绪变化等，是麦麦必须的模型
+model_list = ["siliconflow-deepseek-v3","qwen3-30b"] # 使用的模型列表，每个子项对应上面的模型名称(name)
 temperature = 0.2                        # 模型温度，新V3建议0.1-0.3
 max_tokens = 800                         # 最大输出token数
 
@@ -103,6 +103,11 @@ model_list = ["qwen3-8b","qwen3-30b"]
 temperature = 0.7
 max_tokens = 800
 
+[model_task_config.tool_use] #工具调用模型，需要使用支持工具调用的模型
+model_list = ["qwen3-30b"]
+temperature = 0.7
+max_tokens = 800
+
 [model_task_config.replyer] # 首要回复模型，还用于表达器和表达方式学习
 model_list = ["siliconflow-deepseek-v3"]
 temperature = 0.3                        # 模型温度，新V3建议0.1-0.3
@@ -113,11 +118,6 @@ model_list = ["siliconflow-deepseek-v3"]
 temperature = 0.3
 max_tokens = 800
 
-[model_task_config.emotion] #负责麦麦的情绪变化
-model_list = ["qwen3-30b"]
-temperature = 0.7
-max_tokens = 800
-
 [model_task_config.vlm] # 图像识别模型
 model_list = ["qwen2.5-vl-72b"]
 max_tokens = 800
@@ -125,11 +125,6 @@ max_tokens = 800
 [model_task_config.voice] # 语音识别模型
 model_list = ["sensevoice-small"]
 
-[model_task_config.tool_use] #工具调用模型，需要使用支持工具调用的模型
-model_list = ["qwen3-30b"]
-temperature = 0.7
-max_tokens = 800
-
 #嵌入模型
 [model_task_config.embedding]
 model_list = ["bge-m3"]