Join

Integrations

Integrate your model development pipeline with your favorite machine learning frameworks and libraries, and prepare your data for ingestion into downstream systems. Most of our integrations come in the form of staging functions, which take a list of Element objects as input and return formatted dictionaries as output.

Integration with Argilla

You can convert a list of Text elements to an Argilla Dataset using the stage_for_argilla staging function. Specify the type of dataset to be generated using the argilla_task parameter. Valid values are "text_classification", "token_classification", and "text2text". Follow the link for more details on usage.

Integration with Baseplate

Baseplate is a backend optimized for use with LLMs that has an easy to use spreadsheet interface. The unstructured library offers a staging function to convert a list of Element objects into the rows format required by the Baseplate API. See the stage_for_baseplate documentation for information on how to stage elements for ingestion into Baseplate.

Integration with Datasaur

You can format a list of Text elements as input to token based tasks in Datasaur using the stage_for_datasaur staging function. You will obtain a list of dictionaries indexed by the keys "text" with the content of the element, and "entities" with an empty list. Follow the link to learn how to customise your entities and for more details on usage.

Integration with Hugging Face

You can prepare Text elements for processing in Hugging Face Transformers pipelines by splitting the elements into chunks that fit into the model’s attention window using the stage_for_transformers staging function. You can customise the transformation by defining the buffer and window_size, the split_function and the chunk_separator. if you need to operate on text directly instead of unstructured Text objects, use the chunk_by_attention_window helper function. Follow the links for more details on usage.

Integration with Labelbox

You can format your outputs for use with LabelBox using the stage_for_label_box staging function. LabelBox accepts cloud-hosted data and does not support importing text directly. With this integration you can stage the data files in the output_directory to be uploaded to a cloud storage service (such as S3 buckets) and get a config of type List[Dict[str, Any]] that can be written to a .json file and imported into LabelBox. Follow the link to see how to generate the config.json file that can be used with LabelBox, how to upload the staged data files to an S3 bucket, and for more details on usage.

Integration with Label Studio

You can format your outputs for upload to Label Studio using the stage_for_label_studio staging function. After running stage_for_label_studio, you can write the results to a JSON folder that is ready to be included in a new Label Studio project. You can also include pre-annotations and predictions as part of your upload.

Check our example notebook to format and upload the risk section from an SEC filing to Label Studio for a sentiment analysis labeling task here . Follow the link for more details on usage, and check Label Studio docs for a full list of options for labels and annotations.

Integration with LangChain

Our integration with LangChain makes it incredibly easy to combine language models with your data, no matter what form it is in. The Unstructured.io File Loader extracts the text from a variety of unstructured text files using our unstructured library. It is designed to be used as a way to load data into LangChain. Here is the simplest way to use the UnstructuredFileLoader in langchain.

from langchain.document_loaders import UnstructuredFileLoader

loader = UnstructuredFileLoader("state_of_the_union.txt")
loader.load()

Checkout the LangChain docs for more examples about how to use Unstructured data loaders.

Integration with LlamaIndex

To use Unstructured.io File Loader you will need to have LlamaIndex 🦙 (GPT Index) installed in your environment. Just pip install llama-index and then pass in a Path to a local file. Optionally, you may specify split_documents if you want each element generated by unstructured to be placed in a separate document. Here is a simple example of how to use it:

from pathlib import Path
from llama_index import download_loader


UnstructuredReader = download_loader("UnstructuredReader")

loader = UnstructuredReader()
documents = loader.load_data(file=Path('./10k_filing.html'))

See here for more LlamaHub examples.

Integration with Pandas

You can convert a list of Element objects to a Pandas dataframe with columns for the text from each element and their types such as NarrativeText or Title using the convert_to_dataframe staging function. Follow the link for more details on usage.

Integration with Prodigy

You can format your JSON or CSV outputs for use with Prodigy using the stage_for_prodigy and stage_csv_for_prodigy staging functions. After running stage_for_prodigy | stage_csv_for_prodigy, you can write the results to a .json | .jsonl or a .csv file that is ready to be used with Prodigy. Follow the links for more details on usage.

Integration with Weaviate

Weaviate is an open-source vector database that allows you to store data objects and vector embeddings from a variety of ML models. Storing text and embeddings in a vector database such as Weaviate is a key component of the emerging LLM tech stack. See the stage_for_weaviate docs for details on how to upload unstructured outputs to Weaviate. An example notebook is also available here.