README.md

metadata

dataset_info:
  features:
    - name: text
      dtype: string
    - name: label
      dtype: int64
    - name: label_text
      dtype: string
    - name: idx
      dtype: int64
    - name: query_idx
      dtype: int64
    - name: positive_idx
      dtype: int64
    - name: negative_idx
      dtype: int64
  splits:
    - name: train
      num_bytes: 17791088
      num_examples: 50000
  download_size: 11682866
  dataset_size: 17791088
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

MTEB Toxic Conversations 50k Triplets Dataset

This dataset was used in the paper GISTEmbed: Guided In-sample Selection of Training Negatives for Text Embedding Fine-tuning. Refer to https://arxiv.org/abs/2402.16829 for details.

The code for generating the data is available at https://github.com/avsolatorio/GISTEmbed/blob/main/scripts/create_classification_dataset.py.

Citation

@article{solatorio2024gistembed,
    title={GISTEmbed: Guided In-sample Selection of Training Negatives for Text Embedding Fine-tuning},
    author={Aivin V. Solatorio},
    journal={arXiv preprint arXiv:2402.16829},
    year={2024},
    URL={https://arxiv.org/abs/2402.16829}
    eprint={2402.16829},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}