使用huggingface的text embedding models

python 复制代码
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_community.vectorstores import FAISS
from langchain_community.llms import Tongyi
from langchain_core.runnables import RunnableParallel, RunnablePassthrough
from langchain_community.embeddings import HuggingFaceEmbeddings
import os
import time
os.environ["DASHSCOPE_API_KEY"] = "sk-cc1c8314fdbd43ceaf26ec1824d5dd3b"
llm = Tongyi()

from langchain_community.document_loaders import UnstructuredURLLoader

embeddings = HuggingFaceEmbeddings()

# 记录开始时间
start_time = time.time()
text = "This is a test document."

query_result = embeddings.embed_query(text)

end_time = time.time()
# 计算并打印函数执行时间
execution_time = end_time - start_time
print(f"函数执行时间: {execution_time} 秒")
print(query_result[:3])


urls = [
    "https://en.wikipedia.org/wiki/Android_(operating_system)"
]

loader = UnstructuredURLLoader(urls=urls)
documents = loader.load_and_split()
# print(documents)


# # 第一次存入本地
# vectorstore = FAISS.from_documents(documents, embeddings)
# vectorstore.save_local("faiss_index2")


# 记录开始时间
start_time = time.time()

# # 从本地加载
vectorstore = FAISS.load_local("faiss_index2", embeddings)

retriever = vectorstore.as_retriever()
template = """Answer the question based on the context below. If the
question cannot be answered using the information provided answer
with "I don't know"     

Context: {context}

Question: {question}
"""
prompt = ChatPromptTemplate.from_template(template)

output_parser = StrOutputParser()
setup_and_retrieval = RunnableParallel(
    {"context": retriever, "question": RunnablePassthrough()}
)
chain = setup_and_retrieval | prompt | llm | output_parser
print(chain.invoke("what is android"))
# 计算并打印函数执行时间
end_time = time.time()
execution_time = end_time - start_time
print(f"函数执行时间: {execution_time} 秒")

上面是使用的默认的模型,以下指定使用 all-MiniLM-L6-v2:

python 复制代码
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_community.vectorstores import FAISS
from langchain_community.llms import Tongyi
from langchain_core.runnables import RunnableParallel, RunnablePassthrough
from langchain_community.embeddings import HuggingFaceEmbeddings
import os
import time
os.environ["DASHSCOPE_API_KEY"] = "sk-cc1c8314fdbd43ceaf26ec1824d5dd3b"
llm = Tongyi()

from langchain_community.document_loaders import UnstructuredURLLoader
model_name = "all-MiniLM-L6-v2"
embeddings = HuggingFaceEmbeddings(
         model_name=model_name,
)

# 记录开始时间
start_time = time.time()
text = "This is a test document."

query_result = embeddings.embed_query(text)

end_time = time.time()
# 计算并打印函数执行时间
execution_time = end_time - start_time
print(f"函数执行时间: {execution_time} 秒")
print(query_result[:3])


urls = [
    "https://en.wikipedia.org/wiki/Android_(operating_system)"
]

loader = UnstructuredURLLoader(urls=urls)
documents = loader.load_and_split()
# print(documents)


# 记录开始时间
start_time = time.time()

# 第一次存入本地
vectorstore = FAISS.from_documents(documents, embeddings)
vectorstore.save_local("faiss_index2")


# # 从本地加载
# vectorstore = FAISS.load_local("faiss_index2", embeddings)

retriever = vectorstore.as_retriever()
template = """Answer the question based on the context below. If the
question cannot be answered using the information provided answer
with "I don't know"     

Context: {context}

Question: {question}
"""
prompt = ChatPromptTemplate.from_template(template)

output_parser = StrOutputParser()
setup_and_retrieval = RunnableParallel(
    {"context": retriever, "question": RunnablePassthrough()}
)
chain = setup_and_retrieval | prompt | llm | output_parser
print(chain.invoke("what is android"))
# 计算并打印函数执行时间
end_time = time.time()
execution_time = end_time - start_time
print(f"函数执行时间: {execution_time} 秒")

关于可以使用的模型,可以看这里:

相关推荐
电科_银尘2 天前
【大模型算法与工程 14】Tokenizer与Embedding:文本如何变成模型输入
算法·embedding
LiaoWL1233 天前
从零搭一个 RAG 知识库(二):换上语义 Embedding——BGE 实测与阈值重校准
python·大模型·numpy·embedding·向量检索·rag
LiaoWL1236 天前
NumPy 入门实战(二):向量化与余弦相似度——从 ndarray 到 Embedding
python·numpy·embedding
聪明的笨猪猪8 天前
企业知识库AI架构选型:落地导向的架构决策与组件清单
经验分享·面试·架构·embedding·数据库架构
nice_lcj5209 天前
【从零理解 Spring AI · 14】RAG 的四个环节:加载、切分、向量化、检索
java·spring boot·embedding·向量数据库·rag·spring ai·ai应用开发
民乐团扒谱机18 天前
【一文吃透】机器学习 Embedding 降维打击:从底层空间几何原理到 RAG 向量检索工程实战避坑
人工智能·机器学习·embedding
一木 之林22 天前
Dify学习笔记 04 · 检索底座:Embedding、Rerank与向量数据库(第46-54集)
笔记·学习·embedding·dify
AI小百科25 天前
Embedding 是什么?为什么它是 AI 的“翻译官“
人工智能·embedding
npupengsir1 个月前
LUXICAL文档embedding模型介绍
embedding