README.md

metadata

dataset_info:
  features:
    - name: conversation
      list:
        - name: content
          dtype: string
        - name: role
          dtype: string
  splits:
    - name: train
      num_bytes: 62959
      num_examples: 14
  download_size: 26797
  dataset_size: 62959
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Plans

I'm focusing in datasets that can be useful for my future finetunes.

The goal is to align a model without any censoring.

The following are my 2 small datasets I plan to use for this purpose.

• BornSaint/D33_590d
• BornSaint/D33_alignment: this repo