Upstage

Upstage is a leading artificial intelligence (AI) company specializing in delivering above-human-grade performance LLM components.

Solar Pro is an enterprise-grade LLM optimized for single-GPU deployment, excelling in instruction-following and processing structured formats like HTML and Markdown. It supports English, Korean, and Japanese with top multilingual performance and offers domain expertise in finance, healthcare, and legal.

Other than Solar, Upstage also offers features for real-world RAG (retrieval-augmented generation), such as Document Parse and Groundedness Check.

Upstage LangChain 集成

API	描述	导入	使用示例
Chat	Build assistants using Solar Chat	from langchain_upstage import ChatUpstage	Go
Text Embedding	Embed strings to vectors	from langchain_upstage import UpstageEmbeddings	Go
Groundedness Check	Verify groundedness of assistant's response	from langchain_upstage import UpstageGroundednessCheck	Go
Document Parse	Serialize documents with tables and figures	from langchain_upstage import UpstageDocumentParseLoader	Go

查看更多关于模型和功能的详细信息，请参阅文档。

安装与设置

安装 langchain-upstage 个包：

pip install -qU langchain-core langchain-upstage

获取 API密钥 并设置环境变量 UPSTAGE_API_KEY。

import os

os.environ["UPSTAGE_API_KEY"] = "YOUR_API_KEY"

聊天模型

太阳大型语言模型

查看使用示例。

from langchain_upstage import ChatUpstage

chat = ChatUpstage()
response = chat.invoke("Hello, how are you?")
print(response)

API 参考：ChatUpstage

嵌入模型

查看使用示例。

from langchain_upstage import UpstageEmbeddings

embeddings = UpstageEmbeddings(model="solar-embedding-1-large")
doc_result = embeddings.embed_documents(
    ["Sung is a professor.", "This is another document"]
)
print(doc_result)

query_result = embeddings.embed_query("What does Sung do?")
print(query_result)

API 参考：UpstageEmbeddings

文档加载器

文档解析

查看使用示例。

from langchain_upstage import UpstageDocumentParseLoader

file_path = "/PATH/TO/YOUR/FILE.pdf"
layzer = UpstageDocumentParseLoader(file_path, split="page")

# For improved memory efficiency, consider using the lazy_load method to load documents page by page.
docs = layzer.load()  # or layzer.lazy_load()

for doc in docs[:3]:
    print(doc)

API 参考：UpstageDocumentParseLoader

工具

事实核查

查看使用示例。

from langchain_upstage import UpstageGroundednessCheck

groundedness_check = UpstageGroundednessCheck()

request_input = {
    "context": "Mauna Kea is an inactive volcano on the island of Hawaii. Its peak is 4,207.3 m above sea level, making it the highest point in Hawaii and second-highest peak of an island on Earth.",
    "answer": "Mauna Kea is 5,207.3 meters tall.",
}
response = groundedness_check.invoke(request_input)
print(response)

API 参考：UpstageGroundednessCheck