metadata

license: apache-2.0
extra_gated_fields:
  Name: text
  Affilation: text
  Company: text
  Country: country
  Specific date: date_picker
  I want to use this dataset for:
    type: select
    options:
      - Research
      - Education
      - label: Other
        value: other
  I agree to use this dataset for non-commercial use ONLY: checkbox
task_categories:
  - sentence-similarity
language:
  - ar
tags:
  - STS
  - Embeddings
  - Arabic
pretty_name: Arab3M-Triplets
size_categories:
  - 1M<n<10M

Arab3M-Triplets

This dataset is designed for training and evaluating models using contrastive learning techniques, particularly in the context of natural language understanding. The dataset consists of triplets: an anchor sentence, a positive sentence, and a negative sentence. The goal is to encourage models to learn meaningful representations by distinguishing between semantically similar and dissimilar sentences.

Dataset Overview

Format: Parquet
Number of rows: 3.03 million
Columns:
- anchor: A sentence serving as the reference point.
- positive: A sentence that is semantically similar to the anchor.
- negative: A sentence that is semantically dissimilar to the anchor.

Usage

This dataset can be used to train models for various NLP tasks, including:

Sentence Similarity: Training models to identify sentences with similar meanings.
Contrastive Learning: Teaching models to differentiate between semantically related and unrelated sentences.
Representation Learning: Developing models that learn robust sentence embeddings.

Loading the Dataset

You can load the dataset using the Hugging Face datasets library:

from datasets import load_dataset
dataset = load_dataset('Omartificial-Intelligence-Space/Arab3M-Triplets')