Zero Shot Implicit Bi Encoder

🚀 Zero-shot Implicit Bi-Encoder

A sentence-transformers model for zero-shot text classification, leveraging implicit training and aspect normalization.

🚀 Quick Start

This is a sentence-transformers model introduced in the Findings of ACL'23 Paper Label Agnostic Pre-training for Zero-shot Text Classification by Christopher Clarke, Yuzhao Heng, Yiping Kang, Krisztian Flautner, Lingjia Tang and Jason Mars. The code for training and evaluating this model can be found here.

✨ Features

• Designed for zero-shot text classification.
• Trained under the dual encoding classification framework via implicit training.
• Utilizes the aspect-normalized UTCD dataset.

📦 Installation

No specific installation steps are provided in the original document.

💻 Usage Examples

Basic Usage

>>> from sentence_transformers import SentenceTransformer, util as sbert_util
>>> model = SentenceTransformer(model_name_or_path='claritylab/zero-shot-implicit-bi-encoder')

>>> text = "I'd like to have this track onto my Classical Relaxations playlist."
>>> labels = [
>>>     'Add To Playlist', 'Book Restaurant', 'Get Weather', 'Play Music', 'Rate Book', 'Search Creative Work',
>>>     'Search Screening Event'
>>> ]
>>> aspect = 'intent'
>>> aspect_sep_token = model.tokenizer.additional_special_tokens[0]
>>> text = f'{aspect} {aspect_sep_token} {text}'

>>> text_embed = model.encode(text)
>>> label_embeds = model.encode(labels)
>>> scores = [sbert_util.cos_sim(text_embed, lb_embed).item() for lb_embed in label_embeds]
>>> print(scores)

[
  0.7989747524261475,
  0.003968147560954094,
  0.027803801000118256,
  0.9257574081420898,
  0.1492517590522766,
  0.010640474036335945,
  0.012045462615787983
]

📚 Documentation

Model description

This model is intended for zero-shot text classification. It was trained under the dual encoding classification framework via implicit training with the aspect-normalized UTCD dataset.

• Finetuned from model: bert-base-uncased

📄 License

This project is licensed under the MIT license.