RockyHHH / Safety-EvaluatingLinks
本文提出了一个基于“文心一言”的中国LLMs的安全评估基准，其中包括8种典型的安全场景和6种指令攻击类型。此外，本文还提出了安全评估的框架和过程，利用手动编写和收集开源数据的测试Prompts，以及人工干预结合利用LLM强大的评估能力作为“共同评估者”。
☆31Updated 2 years ago
Alternatives and similar repositories for Safety-Evaluating
Users that are interested in Safety-Evaluating are comparing it to the libraries listed below
Sorting:
- SC-Safety: 中文大模型多轮对抗安全基准☆146Updated last year
- ShieldLM: Empowering LLMs as Aligned, Customizable and Explainable Safety Detectors [EMNLP 2024 Findings]☆214Updated last year
- Official github repo for SafetyBench, a comprehensive benchmark to evaluate LLMs' safety. [ACL 2024]☆260Updated 3 months ago
- Flames is a highly adversarial benchmark in Chinese for LLM's harmlessness evaluation developed by Shanghai AI Lab and Fudan NLP Group.☆60Updated last year
- CCAC2024——大模型安全的双重防线：少样本文本内容安全挑战赛仓库☆28Updated last year
- 基于ChatGPT构建的中文self-instruct数据集☆119Updated 2 years ago
- Dataset and evaluation script for "Evaluating Hallucinations in Chinese Large Language Models"☆135Updated last year
- ☆163Updated 2 years ago
- A Massive Multi-Level Multi-Subject Knowledge Evaluation benchmark☆102Updated 2 years ago
- ☆19Updated last year
- 怎么训练一个LLM分词器☆153Updated 2 years ago
- 专注于中文领域大语言模型，落地到某个行业某个领域，成为一个行业大模型、公司级别或行业级别领域大模型。☆125Updated 7 months ago
- SuperCLUE-Agent: 基于中文原生任务的Agent智能体核心能力测评基准☆93Updated last year
- 中文 Instruction tuning datasets☆138Updated last year
- ☆68Updated 2 years ago
- 一套代码指令微调大模型☆38Updated 2 years ago
- ☆145Updated last year
- Clustering and Ranking: Diversity-preserved Instruction Selection through Expert-aligned Quality Estimation☆90Updated 11 months ago
- ☆147Updated last year
- 面向中文大模型价值观的评估与对齐研究☆542Updated 2 years ago
- ☆98Updated last year
- 使用 Qwen2ForSequenceClassification 简单实现文本分类任务。☆83Updated last year
- ChatGPT相关资源汇总☆56Updated 2 years ago
- ☆29Updated last year
- ☆74Updated 9 months ago
- Official Repository for SIGIR2024 Demo Paper "An Integrated Data Processing Framework for Pretraining Foundation Models"☆84Updated last year
- Official github repo for ACLUE, an evaluation benchmark focused on ancient Chinese language comprehension☆30Updated last year
- llama，chatglm 等模型的微调☆91Updated last year
- MEASURING MASSIVE MULTITASK CHINESE UNDERSTANDING☆89Updated last year
- "他山之石、可以攻玉"：复旦白泽智能发布面向国内开源和国外商用大模型的Demo数据集JADE-DB☆468Updated last week