Raymond102103028/PATCH

PATCH: Prompt Assortment for Traditional Chinese Hazards

The first large-scale adversarial safety dataset for Traditional Chinese (TC), designed to train and evaluate content safety classifiers for lightweight LLMs.

For full documentation, see our GitHub repository.

Dataset Overview

• —593,020 safe prompts localized to Traditional Chinese
• —231,924 unsafe prompts across 13 MLCommons hazard categories
• —PATCH-GPT: Direct harmful prompts
• —PATCH-RT: Evasive prompts exploiting TC-specific cultural/linguistic patterns
• —PATCH-H: Gold-standard human-annotated benchmark (390 prompts, Fleiss' κ = 0.84) — held out from training

Dataset Structure

├── safe/           # PATCH_safe_{train,val,test}.csv
├── unsafe_gpt/     # PATCH_unsafe_gpt_{train,val,test}.csv
└── unsafe_rt/      # PATCH_unsafe_rt_{train,val,test}.csv

All files follow a 70:10:20 train/validation/test split.

Dataset Configurations

The dataset contains three configurations:

• —safe: Safe Traditional Chinese prompts.
• —unsafe_gpt: Direct harmful prompts across MLCommons hazard categories.
• —unsafe_rt: Evasive harmful prompts exploiting Traditional Chinese cultural and linguistic patterns.

Usage

from datasets import load_dataset

safe = load_dataset(
    "Raymond102103028/PATCH",
    "safe"
)

unsafe_gpt = load_dataset(
    "Raymond102103028/PATCH",
    "unsafe_gpt"
)

unsafe_rt = load_dataset(
    "Raymond102103028/PATCH",
    "unsafe_rt"
)

Each configuration contains train, validation, and test splits.

License

MIT License — for research use only.