Add pipeline_tag and paper link to metadata

This PR improves the model card for Hermes 4 (FP8 variant) by adding the following metadata for better discoverability and completeness:

- `pipeline_tag: text-generation`: This accurately reflects the model's primary function as a generative language model capable of diverse text outputs, including reasoning and instruction-following. This will help users find the model in the [text generation models](https://huggingface.co/models?pipeline_tag=text-generation) category.
- `paper: https://huggingface.co/papers/2508.18255`: Links the model directly to its official Hugging Face paper page for easy access to the technical report.

No changes were made to the model card's content, which already includes a Transformers usage example and an in-text link to the paper.

README.md

@@ -1,7 +1,11 @@
 ---
+base_model: meta-llama/Meta-Llama-3.1-405B
 language:
 - en
+library_name: transformers
 license: llama3
+pipeline_tag: text-generation
+paper: https://huggingface.co/papers/2508.18255
 tags:
 - Llama-3.1
 - instruct
@@ -18,18 +22,14 @@ tags:
 - long context
 - roleplaying
 - chat
-base_model: meta-llama/Meta-Llama-3.1-405B
-library_name: transformers
 widget:
 - example_title: Hermes 4
   messages:
   - role: system
-    content: >-
-      You are Hermes 4, a capable, neutrally-aligned assistant. Prefer concise,
+    content: You are Hermes 4, a capable, neutrally-aligned assistant. Prefer concise,
       correct answers.
   - role: user
-    content: >-
-      Explain the difference between BFS and DFS to a new CS student.
+    content: Explain the difference between BFS and DFS to a new CS student.
 model-index:
 - name: Hermes-4-Llama-3.1-405B
   results: []