mdegoede/AI_safety

stars

commits

Python

primary language

May 27, 2025

updated

README

Run main.py to get prompt and response data, or interactive_prompt to create prompt and discussion data.

Label the resulting data by hand (biased/unbiased) in the y column.

Run compare_llm_to_gpt2.py to label the data by gpt-2 and get the accuracy score of gpt-2's labels compared to the labels in the y column.

20 commits

3 commits

1 commits

davidjurgens/human-llm-similarity

TheTerrariaGuy/AI-Bias-Website

An interactive page that shows the output of a NLP model to test for biases in these models.

dplecko/fgai

allenai/real-toxicity-prompts

234

aounon/certified-llm-safety

JoNeedsSleep/team-shard-experiments

sisl/ASTPrompter

Weakly Supervised Automated Language Model Red-Teaming to Identify Likely Toxic Prompts.

paul-rottger/exaggerated-safety

Röttger et al. (NAACL 2024): "XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in…

146

Python

100.0%

README

Run main.py to get prompt and response data, or interactive_prompt to create prompt and discussion data.

Label the resulting data by hand (biased/unbiased) in the y column.

Run compare_llm_to_gpt2.py to label the data by gpt-2 and get the accuracy score of gpt-2's labels compared to the labels in the y column.