调用LLMGraphTransformer.convert_to_graph_documents触发AttributeError错误

在Colab Notebooks环境中，使用HuggingFaceH4/starchat2-15b-v0.1作为大语言模型，执行以下代码调用LLMGraphTransformer.convert_to_graph_documents方法时，触发AttributeError: 'str' object has no attribute 'content'错误：

from langchain_core.documents import Document

text = """
Marie Curie, born in 1867, was a Polish and naturalised-French physicist and chemist who conducted pioneering research on radioactivity.
She was the first woman to win a Nobel Prize, the first person to win a Nobel Prize twice, and the only person to win a Nobel Prize in two scientific fields.
Her husband, Pierre Curie, was a co-winner of her first Nobel Prize, making them the first-ever married couple to win the Nobel Prize and launching the Curie family legacy of five Nobel Prizes.
She was, in 1906, the first woman to become a professor at the University of Paris.
"""
documents = [Document(page_content=text)]
graph_documents = llm_transformer.convert_to_graph_documents(documents)
print(f"Nodes:{graph_documents[0].nodes}")
print(f"Relationships:{graph_documents[0].relationships}")

错误栈信息：

AttributeError                            Traceback (most recent call last)
<ipython-input-38-0a6a66b2d25f> in <cell line: 10>()
      8 """
      9 documents = [Document(page_content=text)]
---> 10 graph_documents = llm_transformer.convert_to_graph_documents(documents)
     11 print(f"Nodes:{graph_documents[0].nodes}")
     12 print(f"Relationships:{graph_documents[0].relationships}")

2 frames
/usr/local/lib/python3.10/dist-packages/langchain_experimental/graph_transformers/llm.py in process_response(self, document)
    593             nodes_set = set()
    594             relationships = []
---> 595             parsed_json = self.json_repair.loads(raw_schema.content)
    596             for rel in parsed_json:
    597                 # Nodes need to be deduplicated using a set

AttributeError: 'str' object has no attribute 'content'

错误原因 #

langchain_experimental的LLMGraphTransformer在处理模型返回结果时，默认假设返回值是带有content属性的对象，但starchat2这类模型直接返回字符串，导致代码尝试访问raw_schema.content时触发属性错误。

解决办法 #

方法1：修改LLMGraphTransformer的process_response方法 #

找到langchain_experimental/graph_transformers/llm.py中的process_response函数，修改返回值处理逻辑，兼容字符串类型：

# 原代码行
parsed_json = self.json_repair.loads(raw_schema.content)
# 修改为
raw_content = raw_schema.content if hasattr(raw_schema, 'content') else raw_schema
parsed_json = self.json_repair.loads(raw_content)

方法2：给LLM添加返回值包装器 #

自定义包装类将字符串返回值转为带content属性的对象，避免修改源码：

class StringWrapper:
    def __init__(self, content):
        self.content = content

# 替换LLM的predict方法
original_predict = llm.predict
def wrapped_predict(prompt):
    result = original_predict(prompt)
    return StringWrapper(result)
llm.predict = wrapped_predict

方法3：升级LangChain版本 #

后续版本的LangChain可能已修复该兼容性问题，执行升级命令：

pip install --upgrade langchain langchain_experimental

内容的提问来源于stack exchange，提问作者Weikun You