traberph/advbench-alpaca-contrastive download history

traberph/advbench-alpaca-contrastive is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 5 times (1 in the last 7 days), and 72 times in total. It ranks #779,199 among datasets by monthly downloads.

AdvBench-Alpaca-Contrastive A contrastive dataset mixing AdvBench prompts for harmful examples and Alpaca prompts for harmless examples. Cached Layers The dataset contains Hidden-state activations extracted from Qwen3-4B-Instruct for probing and activation-steering. For

Open traberph/advbench-alpaca-contrastive on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.