Improve model card with paper, code, project links and pipeline tag

This PR enhances the model card for LLaDA-MoE by adding an explicit link to the foundational paper [Large Language Diffusion Models](https://huggingface.co/papers/2502.09992), the project page (`https://ml-gsai.github.io/LLaDA-demo/`), and the GitHub repository (`https://github.com/ML-GSAI/LLaDA`).

Additionally, it includes the `pipeline_tag: text-generation` in the metadata, improving discoverability on the Hugging Face Hub. The existing sample usage for the `transformers` library confirms its compatibility, and this information is retained.

README.md

@@ -1,14 +1,22 @@
 ---
+library_name: transformers
 license: apache-2.0
 tags:
 - dllm
 - diffusion
 - llm
 - text_generation
-library_name: transformers
+pipeline_tag: text-generation
 ---
+
 # LLaDA-MoE
 
+This model is based on the principles described in the paper [Large Language Diffusion Models](https://huggingface.co/papers/2502.09992).
+
+- 📚 [Paper](https://huggingface.co/papers/2502.09992)
+- 🏠 [Project Page](https://ml-gsai.github.io/LLaDA-demo/)
+- 💻 [Code](https://github.com/ML-GSAI/LLaDA)
+
 **LLaDA-MoE** is a new and upgraded series of the LLaDA diffusion language model. This pre-release includes two cutting-edge models:
 
 - `LLaDA-MoE-7B-A1B-Base`: A base pre-trained model designed for research and secondary development.
@@ -223,10 +231,6 @@ input_ids = torch.tensor(input_ids).to(device).unsqueeze(0)
 
 text = generate(model, input_ids, steps=128, gen_length=128, block_length=32, temperature=0., cfg_scale=0., remasking='low_confidence')
 print(tokenizer.batch_decode(text[:, input_ids.shape[1]:], skip_special_tokens=False)[0])
-
-
-
-
 ```