langchain-upstage 0.2.0

An integration package connecting Upstage and LangChain

langchain-upstage

This package contains the LangChain integrations for Upstage through their APIs.

Installation and Setup

• Install the LangChain partner package

pip install -U langchain-upstage

• Get an Upstage api key from Upstage Console and set it as an environment variable (UPSTAGE_API_KEY)

Chat Models

This package contains the ChatUpstage class, which is the recommended way to interface with Upstage models.

See a usage example

Embeddings

See a usage example

Use solar-embedding-1-large model for embeddings. Do not add suffixes such as -query or -passage to the model name. UpstageEmbeddings will automatically add the suffixes based on the method called.

Layout Analysis Loader

See a usage example

The use_ocr option determines whether OCR will be used for text extraction from documents. If this option is not specified, the default policy of the Upstage Layout Analysis API service will be applied. When use_ocr is set to True, OCR is utilized to extract text. In the case of PDF documents, this involves converting the PDF into images before performing OCR. Conversely, if use_ocr is set to False for PDF documents, the text information embedded within the PDF is used directly. However, if the input document is not a PDF, such as an image, setting use_ocr to False will result in an error.

from langchain_upstage import UpstageLayoutAnalysisLoader

file_path = "/PATH/TO/YOUR/FILE.image"
layzer = UpstageLayoutAnalysisLoader(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)

If you are a Windows user, please ensure that the Visual C++ Redistributable is installed before using the loader.