README.md

metadata

dataset_info:
  features:
    - name: text
      dtype: string
    - name: label
      dtype: string
  splits:
    - name: train
      num_bytes: 559172
      num_examples: 4000
  download_size: 342373
  dataset_size: 559172
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Malaysian-Prompt-Injection

Malaysian style prompt injection dataset, we sampled from,

1. 2000 rows from https://huggingface.co/datasets/mesolitica/Malaysian-Rejection-Prompt
2. 2000 rows from https://huggingface.co/datasets/mesolitica/Malaysian-SFT/blob/6295369f2adf4cf0a2dfcef99e1225f849363e15/combine/combined-malaysian-sft-5k-sample.jsonl