From c53dc6cb69f250deb1663fa17010a2cb8c4a09a1 Mon Sep 17 00:00:00 2001 From: SengokuCola <1026294844@qq.com> Date: Fri, 25 Jul 2025 16:14:41 +0800 Subject: [PATCH] =?UTF-8?q?better=EF=BC=9A=E8=B0=83=E6=95=B4=E6=BF=80?= =?UTF-8?q?=E6=B4=BB=E6=B7=B1=E5=BA=A6?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- .../heart_flow/heartflow_message_processor.py | 1 + src/chat/memory_system/Hippocampus.py | 20 ++++++++++++------- 2 files changed, 14 insertions(+), 7 deletions(-) diff --git a/src/chat/heart_flow/heartflow_message_processor.py b/src/chat/heart_flow/heartflow_message_processor.py index 57b52ae69..6956dbda7 100644 --- a/src/chat/heart_flow/heartflow_message_processor.py +++ b/src/chat/heart_flow/heartflow_message_processor.py @@ -56,6 +56,7 @@ async def _calculate_interest(message: MessageRecv) -> Tuple[float, bool]: with Timer("记忆激活"): interested_rate = await hippocampus_manager.get_activate_from_text( message.processed_plain_text, + max_depth= 5, fast_retrieval=False, ) logger.debug(f"记忆激活率: {interested_rate:.2f}") diff --git a/src/chat/memory_system/Hippocampus.py b/src/chat/memory_system/Hippocampus.py index c1cf61797..c2dfc2180 100644 --- a/src/chat/memory_system/Hippocampus.py +++ b/src/chat/memory_system/Hippocampus.py @@ -224,10 +224,15 @@ class Hippocampus: return hash((source, target)) @staticmethod - def find_topic_llm(text, topic_num): - # sourcery skip: inline-immediately-returned-variable + def find_topic_llm(text:str, topic_num:int|list[int]): + topic_num_str = "" + if isinstance(topic_num, list): + topic_num_str = f"{topic_num[0]}-{topic_num[1]}" + else: + topic_num_str = topic_num + prompt = ( - f"这是一段文字：\n{text}\n\n请你从这段话中总结出最多{topic_num}个关键的概念，可以是名词，动词，或者特定人物，帮我列出来，" + f"这是一段文字：\n{text}\n\n请你从这段话中总结出最多{topic_num_str}个关键的概念，可以是名词，动词，或者特定人物，帮我列出来，" f"将主题用逗号隔开，并加上<>,例如<主题1>,<主题2>......尽可能精简。只需要列举最多{topic_num}个话题就好，不要有序号，不要告诉我其他内容。" f"如果确定找不出主题或者没有明显主题，返回。" ) @@ -325,12 +330,13 @@ class Hippocampus: else: # 使用LLM提取关键词 - 根据详细文本长度分布优化topic_num计算 text_length = len(text) + topic_num:str|list[int] = None if text_length <= 5: - topic_num = 1 # 1-5字符: 1个关键词 (26.57%的文本) + topic_num = [1,2] # 1-5字符: 1个关键词 (26.57%的文本) elif text_length <= 10: - topic_num = 1 # 6-10字符: 1个关键词 (27.18%的文本) + topic_num = 2 # 6-10字符: 1个关键词 (27.18%的文本) elif text_length <= 20: - topic_num = 2 # 11-20字符: 2个关键词 (22.76%的文本) + topic_num = [2,3] # 11-20字符: 2个关键词 (22.76%的文本) elif text_length <= 30: topic_num = 3 # 21-30字符: 3个关键词 (10.33%的文本) elif text_length <= 50: @@ -721,7 +727,7 @@ class Hippocampus: for keyword in valid_keywords: logger.debug(f"开始以关键词 '{keyword}' 为中心进行扩散检索 (最大深度: {max_depth}):") # 初始化激活值 - activation_values = {keyword: 1.0} + activation_values = {keyword: 1.5} # 记录已访问的节点 visited_nodes = {keyword} # 待处理的节点队列，每个元素是(节点, 激活值, 当前深度)