Add link to paper

README.md

@@ -1,16 +1,16 @@
 ---
+language:
+- en
 license: apache-2.0
+size_categories:
+- 1K<n<10K
 task_categories:
 - question-answering
-language:
-- en
 tags:
 - Reasoning
 - LLM
 - Encryption
 - Decryption
-size_categories:
-- 1K<n<10K
 configs:
 - config_name: Rot13
   data_files:
@@ -49,6 +49,7 @@ configs:
   - split: test
     path: data/WordShift.jsonl
 ---
+
 # CipherBank Benchmark
 
 ## Benchmark description
@@ -76,6 +77,7 @@ We evaluate state-of-the-art LLMs on CipherBank, e.g., GPT-4o, DeepSeek-V3, and
 |**o1-2024-12-17** | **40.59**   |
 
 ## Please see paper & website for more information:
+- [Paper](https://huggingface.co/papers/2504.19093)
 - [https://arxiv.org/abs/2504.19093](https://arxiv.org/abs/2504.19093)
 - [https://cipherbankeva.github.io/](https://cipherbankeva.github.io/)
 
@@ -91,4 +93,4 @@ If you find CipherBank useful for your research and applications, please cite us
       primaryClass={cs.CR},
       url={https://arxiv.org/abs/2504.19093}, 
 }
-```
+```