NIM 平台生成式 AI-demo

~~需要python环境~~

官网注册：（后续调用模型需要秘钥key）Try NVIDIA NIM APIs

可以看到有多种模型：

官方案例

1.安装相关依赖：

pip install langchain_nvidia_ai_endpoints langchain-community langchain-text-splitters faiss-cpu gradio==3.50.0 setuptools beautifulsoup4

2.官方案例：

# -*- coding: utf-8 -*-# 导入必要的库
from langchain_nvidia_ai_endpoints import NVIDIAEmbeddings, ChatNVIDIA
from langchain_community.document_loaders import WebBaseLoader
from langchain_community.vectorstores import FAISS
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import chain
import os
import gradio as gr
from datetime import datetime
# Even if you do not know the full answer, generate a one-paragraph hypothetical answer to the below question in Chinese
# 定义假设性回答模板
hyde_template = """Even if you do not know the full answer, generate a one-paragraph hypothetical answer to the below question:{question}"""# 定义最终回答模板
template = """Answer the question based only on the following context:
{context}Question: {question}
"""# 定义函数来处理问题
def process_question(url, api_key, model_name, question):# 初始化加载器并加载数据loader = WebBaseLoader(url)docs = loader.load()# 设置环境变量os.environ['NVIDIA_API_KEY'] = api_key# 初始化嵌入层embeddings = NVIDIAEmbeddings()text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)documents = text_splitter.split_documents(docs)vector = FAISS.from_documents(documents, embeddings)retriever = vector.as_retriever()# 初始化模型model = ChatNVIDIA(model=model_name)# 创建提示模板hyde_prompt = ChatPromptTemplate.from_template(hyde_template)hyde_query_transformer = hyde_prompt | model | StrOutputParser()# 定义检索函数@chaindef hyde_retriever(question):hypothetical_document = hyde_query_transformer.invoke({"question": question})return retriever.invoke(hypothetical_document)# 定义最终回答链prompt = ChatPromptTemplate.from_template(template)answer_chain = prompt | model | StrOutputParser()@chaindef final_chain(question):documents = hyde_retriever.invoke(question)response = ""for s in answer_chain.stream({"question": question, "context": documents}):response += sreturn response# 调用最终链获取答案return str(datetime.now())+final_chain.invoke(question)# 定义可用的模型列表
models = ["mistralai/mistral-7b-instruct-v0.2","meta/llama-3.1-405b-instruct"]# 启动Gradio应用
iface = gr.Interface(fn=process_question,inputs=[gr.Textbox(label="输入需要学习的网址"),gr.Textbox(label="NVIDIA API Key"),gr.Dropdown(models, label="选择语言模型"),gr.Textbox(label="输入问题")],outputs="text",title="网页知识问答系统"
)# 启动Gradio界面
iface.launch()