Zero-shot explicit bi-encoder open-source text classification model - Complete text classification quickly without samples

Home

Zero Shot Explicit Bi Encoder

Developed by claritylab

A BERT-based zero-shot text classification model trained on the UTCD dataset using explicit training methods

Text Classification

Transformers

EnglishOpen Source License:MIT #Zero-shot classification #Dual-encoder architecture #Text intent recognition

Downloads 31

Release Time : 5/15/2023

Model Overview

A dual-encoder model specifically designed for zero-shot text classification tasks, optimized for label-agnostic pre-training through aspect-normalized processing

Model Features

Zero-shot learning capability

Classify new categories without domain-specific training data

Explicit training framework

Optimizes classification performance through label-agnostic pre-training methods

Dual-encoder architecture

Independent encoding of text and labels enables efficient similarity calculation

Model Capabilities

Zero-shot text classification

Semantic similarity calculation

Multi-label classification

Use Cases

Intelligent assistants

User intent recognition

Identify potential intent categories in user queries

Accurately distinguishes similar intents like play music/add to playlist

Content classification

Dynamic labeling system

Automatically assign labels to unseen new content

🚀 Zero-shot Explicit Bi-Encoder

A sentence-transformers model for zero-shot text classification, introduced in ACL'23 findings.

🚀 Quick Start

This is a sentence-transformers model. It was 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

Zero-shot Text Classification: Designed specifically for zero-shot text classification tasks.
Trained on UTCD Dataset: Utilizes the aspect-normalized UTCD dataset for explicit training under the dual encoding classification framework.
Finetuned from BERT: Finetuned from the well-known bert-base-uncased model.

📦 Installation

No specific installation steps are provided in the original README. If you want to use the model, you can install the sentence-transformers library:

pip install sentence-transformers

💻 Usage Examples

Basic Usage

>>> from sentence_transformers import SentenceTransformer, util as sbert_util
>>> model = SentenceTransformer(model_name_or_path='claritylab/zero-shot-explicit-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'
>>> ]

>>> 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)

[
  0.53502357006073,
  0.051911696791648865,
  0.0546676367521286,
  0.5633962750434875,
  0.28765711188316345,
  0.17751818895339966,
  0.18489906191825867
]

📚 Documentation

Model description

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

Finetuned from model: bert-base-uncased

📄 License

This project is licensed under the MIT license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご