This is one of the most powerful and “magical” tasks in modern AI.
- Sentiment Analysis is trained on “positive” and “negative.”
- Multi-Label Classification is trained on “toxic,” “insult,” etc.
What if you want to classify a news article into “Politics,” “Business,” or “Sports” without training a new model? That’s Zero-Shot Classification.
Step 1: Install
pip install transformers torch
Step 2: The Code
You just give the pipeline a list of “candidate labels” that you want it to choose from.
from transformers import pipeline
# 1. Load the pipeline
# This will download a model trained on NLI (Natural Language Inference)
classifier = pipeline("zero-shot-classification")
# 2. Define your text and your custom labels
text = "The government announced a new tax policy today that will affect small businesses."
my_labels = ["Sports", "Technology", "Politics", "Business"]
# 3. Classify!
result = classifier(text, candidate_labels=my_labels)
print(result)Step 3: The Result
The model will return a list of your labels, sorted by how relevant they are to the text.
{
'sequence': 'The government announced a new tax policy today that will affect small businesses.',
'labels': ['Politics', 'Business', 'Technology', 'Sports'],
'scores': [0.95, 0.88, 0.02, 0.01]
}The model correctly identified the text is about “Politics” and “Business,” even though it was never specifically trained on those words! You can change the my_labels list to anything (e.g., ["Happy", "Angry", "Urgent"]) and it will work.
Key Takeaways
- Sentiment Analysis and Multi-Label Classification are popular tasks in AI, trained on specific sentiments or categories.
- Zero-Shot Classification allows you to classify text into categories like ‘Politics’, ‘Business’, or ‘Sports’ without training a new model.
- To use Zero-Shot Classification, install the necessary tools and provide a list of candidate labels for the model to choose from.
- The model returns labels sorted by relevance, demonstrating its ability to generalise to unseen categories.
