Add paper link and ecosystem metadata

Hi! I'm Niels from the community science team at Hugging Face.

This PR improves the model card for ROME-30B-A3B by adding the `arxiv` ID to the metadata, which links the model to its technical report on the Hugging Face Hub. I've also added descriptive tags (`agent`, `moe`) and linked the related GitHub repositories for the **ROCK** and **ROLL** components of the Agentic Learning Ecosystem (ALE) mentioned in the paper.

These changes help users discover the model through research and navigate the associated tooling.

README.md

@@ -1,6 +1,10 @@
 ---
 license: apache-2.0
 pipeline_tag: text-generation
+arxiv: 2512.24873
+tags:
+- agent
+- moe
 ---
 
 # ROME-30B-A3B (Coming Soon)
@@ -35,9 +39,9 @@ Rather than scaling performance purely by increasing parameter count, ROME achie
 
 
 ### 🔧 ALE Full-Stack Infrastructure
-- **ROLL** – Large-scale reinforcement learning optimization engine  
+- [**ROLL**](https://github.com/alibaba/ROLL) – Large-scale reinforcement learning optimization engine  
 
-- **ROCK** – Secure sandbox and environment orchestration for agent execution  
+- [**ROCK**](https://github.com/alibaba/ROCK) – Secure sandbox and environment orchestration for agent execution  
 
 - **iFlow CLI** – Unified agent framework and developer interface  
 
@@ -103,4 +107,5 @@ If you find our work useful, please consider citing:
   author={Wang, Weixun and Xu, XiaoXiao and An, Wanhe and Dai, Fangwen and others},
   journal={arXiv preprint arXiv:2512.24873},
   year={2025}
-}
+}
+```