Amazon Press Release Paragraph Classifier

<!-- Provide a quick summary of what the model is/does. -->

The amz-press-release-paragraph-classifier performs multiclass text classification task. Specifically, it takes a blob of text and attributes it to one or more labels. Each of the labels corresponds to a specific section of a press release.

• summary
• opportunity_problem
• approach_solution
• leader_quote
• customer_experience
• customer_testimonial
• call_to_action

Usage

Pipeline

from transformers import pipeline

classification_task = pipeline("text-classification",
  model=model_path,
  tokenizer=model_path,
  top_k=None
)

classification_task('''
   The launch of the AWS Asia Pacific (Melbourne) Region will enable local customers with
   data residency preferences to store data securely in Australia while providing customers
   with even lower latency to drive greater productivity, more efficient business operations,
   and enhanced real-time application performance. Customers will also have access to advanced
   AWS technologies to drive innovation including compute, storage, networking, business
   applications, developer tools, data analytics, security, machine learning, and artificial
   intelligence.
''')

The output:

[
  [
    {'label': 'approach_solution', 'score': 0.3786},
    {'label': 'leader_quote', 'score': 0.1208},
    {'label': 'customer_experience', 'score': 0.9291}
  ]
]