Update dataset card: arXiv 2610.29225 link and citation, dashboard link

README.md

@@ -9,6 +9,7 @@ task_categories:
 - text-generation
 pretty_name: DiagSafe
 tags:
+- arxiv:2610.29225
 - safety
 - llm-safety
 - red-teaming
@@ -40,7 +41,7 @@ configs:
 
 **Beyond Binary Labels: Diagnostic Evaluation of LLM Safety with Prompt-Level Rubrics**
 
-[GitHub](https://github.com/Tencent-Hunyuan/DiagSafe) · [Paper (PDF)](https://github.com/Tencent-Hunyuan/DiagSafe/blob/main/paper/DiagSafe.pdf) · [Evaluation code](https://github.com/Tencent-Hunyuan/DiagSafe#quick-start)
+[Paper (arXiv)](https://arxiv.org/abs/2610.29225) · [GitHub](https://github.com/Tencent-Hunyuan/DiagSafe) · [Evaluation code](https://github.com/Tencent-Hunyuan/DiagSafe#quick-start) · [Dashboard](https://tencent-hunyuan.github.io/DiagSafe/)
 
 > **Content warning.** This dataset contains harmful requests (e.g. violence, sexual content, self-harm,
 > child safety, CBRN) written to test LLM safety. It is intended for safety evaluation and research only.
@@ -155,7 +156,7 @@ multi-turn judging, checkpointing and grouped statistics.
 ## Intended Use and Limitations
 
 DiagSafe is intended for evaluating and diagnosing the safety behavior of LLMs and LLM judges. The prompts
-are deliberately harmful and must not be used to train models to produce harmful content. Results depend
+are deliberately harmful; we ask users not to use them to train models to produce harmful content. Results depend
 on the judge model; the paper reports rubric-based judging at 89.41%–95.34% agreement with human
 annotations across judges.
 
@@ -169,10 +170,13 @@ Copyright (C) 2026 Tencent. Released under the
 ```bibtex
 @misc{diagsafe2026,
   title        = {Beyond Binary Labels: Diagnostic Evaluation of LLM Safety with Prompt-Level Rubrics},
-  author       = {[Author One] and [Author Two] and [Author Three] and [Author Four]},
+  author       = {Li, Hongyu and Huang, Xinlong and Yu, Zhan and Zhao, Jing and Zhao, Zhe and
+                  Zhong, Juntao and Kuang, Xiong and Yuan, Rui and Yan, ZhenXiang and Xu, Huilin and
+                  Yu, Haitao and Wang, Lilin and Chen, Rui and Chen, Blaze and Lian, Fengzong and
+                  Hu, Renfen},
   year         = {2026},
   archivePrefix= {arXiv},
-  eprint       = {XXXX.XXXXX},
+  eprint       = {2610.29225},
   primaryClass = {cs.CL}
 }
 ```