Summary

The Summary pipeline summarizes text. This pipeline runs a text2text model that abstractively creates a summary of the input text.

Example

The following shows a simple example using this pipeline.

from txtai.pipeline import Summary

# Create and run pipeline
summary = Summary()
summary("Enter long, detailed text to summarize here")

See the link below for a more detailed example.

Notebook	Description
Building abstractive text summaries	Run abstractive text summarization

Configuration-driven example

Pipelines are run with Python or configuration. Pipelines can be instantiated in configuration using the lower case name of the pipeline. Configuration-driven pipelines are run with workflows or the API.

config.yml

# Create pipeline using lower case class name
summary:

# Run pipeline with workflow
workflow:
  summary:
    tasks:
      - action: summary

Run with Workflows

from txtai import Application

# Create and run pipeline with workflow
app = Application("config.yml")
list(app.workflow("summary", ["Enter long, detailed text to summarize here"]))

Run with API

CONFIG=config.yml uvicorn "txtai.api:app" &

curl \
  -X POST "http://localhost:8000/workflow" \
  -H "Content-Type: application/json" \
  -d '{"name":"summary", "elements":["Enter long, detailed text to summarize here"]}'

Methods

Python documentation for the pipeline.

__init__(path=None, quantize=False, gpu=True, batch=64, **kwargs)

Source code in txtai/pipeline/text/summary.py

def __init__(self, path=None, quantize=False, gpu=True, batch=64, **kwargs):
    # Default model
    path = path if path else "sshleifer/distilbart-cnn-12-6"

    # Call parent constructor
    super().__init__(path, quantize, gpu, batch)

    # Load model and tokenizer
    self.model, self.tokenizer = self.load(path, "summarization", **kwargs)

__call__(text, minlength=None, maxlength=None, **kwargs)

Runs a summarization model against a block of text.

This method supports text as a string or a list. If the input is a string, the return type is text. If text is a list, a list of text is returned with a row per block of text.

Parameters:

Name	Type	Description	Default
text		text|list	required
minlength		minimum length for summary	None
maxlength		maximum length for summary	None
kwargs		additional keyword arguments	{}

Returns:

Type	Description
	summary text

Source code in txtai/pipeline/text/summary.py

def __call__(self, text, minlength=None, maxlength=None, **kwargs):
    """
    Runs a summarization model against a block of text.

    This method supports text as a string or a list. If the input is a string, the return
    type is text. If text is a list, a list of text is returned with a row per block of text.

    Args:
        text: text|list
        minlength: minimum length for summary
        maxlength: maximum length for summary
        kwargs: additional keyword arguments

    Returns:
        summary text
    """

    # Validate text length greater than max length
    check = maxlength if maxlength else self.maxlength()

    # Skip text shorter than max length
    texts = text if isinstance(text, list) else [text]
    params = [(x, text if len(text) >= check else None) for x, text in enumerate(texts)]

    # Build keyword arguments
    kwargs = self.args(minlength, maxlength)

    inputs = [text for _, text in params if text]
    if inputs:
        # Run summarization
        results = self.generate(inputs, **kwargs)

        # Pull out summary text
        results = iter([self.clean(x) for x in results])
        results = [next(results) if text else texts[x] for x, text in params]
    else:
        # Return original
        results = texts

    return results[0] if isinstance(text, str) else results