NLP / Zero-Shot Classification
Classify Anything
Zero-shot classification for English text. Define your own classes, paste an excerpt, get predictions. All processing happens locally in your browser.
Labels
Create the classes your prompt should be sorted into.
Labels compete for one best match; scores are relative to this label set. Scores are model estimates, not measured accuracy.
Define the classes you want to classify text into
Input Text
Paste an English sentence, paragraph, or short document excerpt.
Results
Review the predicted label distribution and model status.
Enter text and click Classify to see predictions
How it works
1. Define Classes
Add the labels you want to classify text into. Use presets for common tasks like sentiment analysis or topic classification.
2. Enter Text
Use English sentences or short paragraphs. The model reads up to 512 tokens, including the label hypothesis and special tokens. Longer inputs are automatically truncated; their full content is not classified. Split long documents into excerpts.
3. Get Predictions
The model predicts which class best matches your text, with confidence scores for each label. No data leaves your browser.
Using MobileBERT-uncased-MNLI via Transformers.js. Model runs entirely in your browser using WebAssembly.