DeepSeek-3.1

The 'DeepSeek-3.1' dataset/benchmark is used to evaluate the effectiveness of safety alignment in Large Language Models (LLMs) by providing a structured framework for testing the models' responses to potentially harmful prompts.

Papers using DeepSeek-3.1 (1)

The Trojan Example: Jailbreaking LLMs through Template Filling and Unsafety Reasoning2025