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
安替-AnTi16 小时前
PandaWiki:AI 驱动的开源知识库系
人工智能·embedding·检索增强·知识库·rag·查询优化
ghostwritten1 天前
深入理解嵌入模型(Embedding Model):AI 语义世界的基石
人工智能·embedding
Paraverse_徐志斌3 天前
RAG架构(检索增强生成)与向量数据库
数据库·ai·llm·embedding·milvus·rag
AIzealot无3 天前
Qwen3 Embedding报告随笔
人工智能·深度学习·算法·论文·embedding·论文笔记·搜广推
喜欢吃豆9 天前
从潜在空间到实际应用:Embedding模型架构与训练范式的综合解析
python·自然语言处理·架构·大模型·微调·embedding
猫头虎10 天前
HAMi 2.7.0 发布:全面拓展异构芯片支持,优化GPU资源调度与智能管理
嵌入式硬件·算法·prompt·aigc·embedding·gpu算力·ai-native
fyakm18 天前
词向量:从 One-Hot 到 BERT Embedding,NLP 文本表示的核心技术
自然语言处理·bert·embedding
云帆@22 天前
qwen3_embedding
embedding
人工智能培训23 天前
Transformer-位置编码(Position Embedding)
人工智能·深度学习·大模型·transformer·embedding·vision
luoganttcc1 个月前
PyTorch 中nn.Embedding
pytorch·深度学习·embedding