Streamlit集成GPT4All时出现LLMChain参数验证错误求助

尝试将GPT4All与Streamlit结合开发可与上传PDF对话的应用，但运行时触发参数值异常，报错信息如下：

2023-08-24 18:41:50.816 Uncaught app exception
Traceback (most recent call last):
  File "C:\Users\MudassarMa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 556, in _run_script
    exec(code, module.__dict__)
  File "C:\Users\MudassarMa\Downloads\Misc\DataScience\taxgpt\main.py", line 153, in <module>
    main()
  File "C:\Users\MudassarMa\Downloads\Misc\DataScience\taxgpt\main.py", line 148, in main
    st.session_state.conversation = get_conversation_chain(
                                    ^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\MudassarMa\Downloads\Misc\DataScience\taxgpt\main.py", line 85, in get_conversation_chain
    conversation_chain = ConversationalRetrievalChain.from_llm(
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\MudassarMa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\chains\conversational_retrieval\base.py", line 213, in from_llm
    doc_chain = load_qa_chain(
                ^^^^^^^^^^^^^^
  File "C:\Users\MudassarMa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\chains\question_answering\__init__.py", line 238, in load_qa_chain
    return loader_mapping[chain_type](
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\MudassarMa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\chains\question_answering\__init__.py", line 70, in _load_stuff_chain
    llm_chain = LLMChain(
                ^^^^^^^^^
  File "C:\Users\MudassarMa\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.11_qbz5n2kfra8p0\LocalCache\local-packages\Python311\site-packages\langchain\load\serializable.py", line 61, in __init__
    super().__init__(**kwargs)
  File "pydantic\main.py", line 341, in pydantic.main.BaseModel.__init__
pydantic.error_wrappers.ValidationError: 1 validation error for LLMChain
llm
  value is not a valid dict (type=type_error.dict)

报错原因 #

直接使用了gpt4all.gpt4all.GPT4All的原生实例，而LangChain的ConversationalRetrievalChain要求传入的是LangChain封装的LLM类实例，原生GPT4All对象与LangChain接口不兼容，导致参数验证失败。

解决方案 #

1. 替换LLM导入语句 #

将原生GPT4All的导入替换为LangChain封装的版本：

# 替换原来的 from gpt4all.gpt4all import GPT4All
from langchain.llms import GPT4All

2. 修改LLM初始化代码 #

在get_conversation_chain函数中，用LangChain的GPT4All类初始化模型，确保参数符合LangChain要求：

def get_conversation_chain(vector_store):
    # 使用LangChain封装的GPT4All类
    llm = GPT4All(
        model="./ggml-gpt4all-j-v1.3-groovy.bin",  # 确保模型文件路径正确
        verbose=True,  # 可选：开启日志便于调试
        temperature=0.7  # 可选：调整生成温度
    )

    memory = ConversationBufferMemory(
        memory_key='chat_history', return_messages=True)

    conversation_chain = ConversationalRetrievalChain.from_llm(
        llm=llm,
        retriever=vector_store.as_retriever(),
        memory=memory
    )

    return conversation_chain

3. 验证模型路径 #

确保ggml-gpt4all-j-v1.3-groovy.bin文件在指定路径下（当前目录./），如果模型文件在其他位置，需修改model参数的路径。

完整修改后的核心代码片段 #

# 导入部分修正
from langchain import HuggingFaceHub, LLMChain, PromptTemplate
import streamlit as st
from dotenv import load_dotenv
from PyPDF2 import PdfReader
from langchain.text_splitter import CharacterTextSplitter
from langchain.embeddings import OpenAIEmbeddings, HuggingFaceInstructEmbeddings
from langchain.vectorstores import FAISS
from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationalRetrievalChain
from langchain.chat_models import ChatOpenAI
from htmlTemplates import bot_template, user_template, css
import transformers
from transformers import pipeline
from langchain.llms import GPT4All  # 替换原生导入

# ... 其他函数不变，只修改get_conversation_chain ...

def get_conversation_chain(vector_store):
    llm = GPT4All(
        model="./ggml-gpt4all-j-v1.3-groovy.bin",
        verbose=True,
        temperature=0.7
    )

    memory = ConversationBufferMemory(
        memory_key='chat_history', return_messages=True)

    conversation_chain = ConversationalRetrievalChain.from_llm(
        llm=llm,
        retriever=vector_store.as_retriever(),
        memory=memory
    )

    return conversation_chain

# ... 其余代码不变 ...

内容的提问来源于stack exchange，提问作者user810258