RockyHHH / Safety-Evaluating
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”。
☆19Updated last year
Related projects: ⓘ
- SC-Safety: 中文大模型多轮对抗安全基准☆94Updated 6 months ago
- 一套代码指令微调大模型☆36Updated last year
- 基于ChatGPT构建的中文self-instruct数据集☆110Updated last year
- make LLM easier to use☆59Updated last year
- Source code for ACL 2023 paper Decoder Tuning: Efficient Language Understanding as Decoding☆47Updated last year
- Flames is a highly adversarial benchmark in Chinese for LLM's harmlessness evaluation developed by Shanghai AI Lab and Fudan NLP Group.☆30Updated 4 months ago
- llama，chatglm 等模型的微调☆79Updated 2 months ago
- Official github repo for SafetyBench, a comprehensive benchmark to evaluate LLMs' safety.☆141Updated 2 months ago
- ☆124Updated 2 months ago
- Llama2-SFT, Llama-2-7B微调(transformers)/LORA(peft)/推理☆20Updated last year
- 基于 LoRA 和 P-Tuning v2 的 ChatGLM-6B 高效参数微调☆54Updated last year
- ☆23Updated last year
- 用于AIOPS24挑战赛的Demo☆53Updated 3 months ago
- 百川Dynamic NTK-ALiBi的代码实现：无需微调即可推理更长文本☆45Updated last year
- ☆20Updated 2 months ago
- 1.4B sLLM for Chinese and English - HammerLLM🔨☆43Updated 5 months ago
- (NBCE)Naive Bayes-based Context Extension on ChatGLM-6b☆14Updated last year
- 阿里天池: 2023全球智能汽车AI挑战赛——赛道一：AI大模型检索问答 baseline 80+☆63Updated 8 months ago
- Silk Road will be the dataset zoo for Luotuo(骆驼). Luotuo is an open sourced Chinese-LLM project founded by 陈启源 @ 华中师范大学 & 李鲁鲁 @ 商汤科技 & 冷子…☆38Updated 10 months ago
- ☆90Updated 6 months ago
- Clustering and Ranking: Diversity-preserved Instruction Selection through Expert-aligned Quality Estimation☆61Updated 4 months ago
- Dataset and evaluation script for "Evaluating Hallucinations in Chinese Large Language Models"☆105Updated 3 months ago
- “悟道”数据☆39Updated 3 years ago
- Chinese Generation Evaluation☆12Updated last year
- ☆46Updated 2 months ago
- ☆89Updated 9 months ago
- 大型语言模型实战指南：应用实践与场景落地☆23Updated last week
- A repo for update and debug Mixtral-7x8B、MOE、ChatGLM3、LLaMa2、 BaChuan、Qwen an other LLM models include new models mixtral, mixtral 8x7b, …☆40Updated 5 months ago
- GoGPT中文指令数据集构造☆10Updated 7 months ago
- ChatGLM2-6B微调, SFT/LoRA, instruction finetune☆107Updated last year