Adaptive Stress Testing for Language Model Toxicity
Failed to add items
Sorry, we are unable to add the item because your shopping basket is already at capacity.
Add to cart failed.
Please try again later
Add to wishlist failed.
Please try again later
Remove from wishlist failed.
Please try again later
Follow podcast failed
Unfollow podcast failed
Adaptive Stress Testing for Language Model Toxicity
-
Narrated by:
-
Written by:
About this listen
This episode explores ASTPrompter, a novel approach to automated red-teaming for large language models (LLMs). Unlike traditional methods that focus on simply triggering toxic outputs, ASTPrompter is designed to discover likely toxic prompts – those that could naturally emerge during regular language model use. The approach uses Adaptive Stress Testing (AST), a technique that identifies likely failure points, and reinforcement learning to train an "adversary" model. This adversary generates prompts that aim to elicit toxic responses from a "defender" model, but importantly, these prompts have a low perplexity, meaning they are realistic and likely to occur, unlike many prompts generated by other methods.
No reviews yet