docs: publish KernelGenBench arXiv link (#62)

Co-authored-by: automerge-bot <devnull@local>

Author: factnn

README.md

@@ -2,6 +2,8 @@
 
 [[中文版](./README.zh-CN.md)|English]
 
+> **Paper**: [KernelGenBench: A Multi-Source and Multi-Chip Benchmark for LLM-based Kernel Generation](https://arxiv.org/abs/2607.27231) (arXiv:2607.27231)
+
 <div align="right">
   <a href="https://www.linkedin.com/company/flagos-community" target="_blank">
     <img src="./docs/assets/Linkedin.png" alt="LinkedIn" width="32" height="32" />
@@ -26,8 +28,6 @@ KernelGenBench is a component of [FlagOS](https://flagos.io/) — a unified, ope
 
 KernelGenBench is a benchmark framework for evaluating LLM and agent-based Triton kernel generation across multiple hardware platforms.
 
-> **Paper**: [KernelGenBench: A Multi-Source and Multi-Chip Benchmark for LLM-based Kernel Generation](paper/KernelGenBench_paper.pdf) (Under Review)
-
 ![KernelGenBench Overview](docs/source/_static/images/figures/overview.png)
 
 ## Features
@@ -78,13 +78,12 @@ python scripts/generate_kernel_and_verify.py \
 ## Citation
 
 ```bibtex
-@inproceedings{kernelgenbench2026,
+@article{zang2026kernelgenbench,
   title     = {KernelGenBench: A Multi-Source and Multi-Chip Benchmark for LLM-based Kernel Generation},
-  author    = {Anonymous Author(s)},
-  booktitle = {Under review},
+  author    = {Peiyu Zang and Jian Tao and Jialing Zhang and Yichen Yuan and Wentao Zhang and Guang Liu and Yonghua Lin},
+  journal   = {arXiv preprint arXiv:2607.27231},
   year      = {2026},
-  note      = {Preprint PDF and repository available at \url{https://github.com/flagos-ai/KernelGenBench}},
-  url       = {https://github.com/flagos-ai/KernelGenBench}
+  url       = {https://arxiv.org/abs/2607.27231}
 }
 ```
 

README.zh-CN.md

@@ -18,7 +18,7 @@
 
 跨多种硬件平台评测 LLM 和 Agent 生成 Triton kernel 能力的基准框架。
 
-> **论文**: [KernelGenBench: A Multi-Source and Multi-Chip Benchmark for LLM-based Kernel Generation](paper/KernelGenBench_paper.pdf) (Under Review)
+> **论文**：[KernelGenBench: A Multi-Source and Multi-Chip Benchmark for LLM-based Kernel Generation](https://arxiv.org/abs/2607.27231)（arXiv:2607.27231）
 
 ![KernelGenBench 架构概览](docs/source/_static/images/figures/overview.png)
 
@@ -334,13 +334,12 @@ KernelGenBench 是 [FlagOS](https://github.com/flagos-ai) 开源生态的一部
 如果 KernelGenBench 对您的研究或评测有帮助，请引用：
 
 ```bibtex
-@inproceedings{kernelgenbench2026,
+@article{zang2026kernelgenbench,
   title     = {KernelGenBench: A Multi-Source and Multi-Chip Benchmark for LLM-based Kernel Generation},
-  author    = {Anonymous Author(s)},
-  booktitle = {Under review},
+  author    = {Peiyu Zang and Jian Tao and Jialing Zhang and Yichen Yuan and Wentao Zhang and Guang Liu and Yonghua Lin},
+  journal   = {arXiv preprint arXiv:2607.27231},
   year      = {2026},
-  note      = {Preprint PDF and repository available at \url{https://github.com/flagos-ai/KernelGenBench}},
-  url       = {https://github.com/flagos-ai/KernelGenBench}
+  url       = {https://arxiv.org/abs/2607.27231}
 }
 ```