List your company

ResearchFormerly WizWand

LLaMA Chatbot

Benchmarks

TaskDatasetSOTA resultTrendResults

Adversarial Toxicity Refusal

LLaMA-2 Chatbot Specialized category

59.8Refusal Rate (RTR)

3

Adversarial Toxicity Refusal

LLaMA-2 Chatbot Offensive category

47.7RTR

3