Add pipeline and OCR tags, update paper link

This PR improves the model card by:
- Adding the `pipeline_tag: image-to-text` to the metadata, ensuring the model is discoverable under the correct task on the Hugging Face Hub.
- Adding `tags: - ocr` to further enhance discoverability for OCR-specific use cases.
- Updating the paper link in the "Quick links" section to point to the official Hugging Face paper page for [olmOCR 2: Unit Test Rewards for Document OCR](https://huggingface.co/papers/2510.19817).

README.md

@@ -1,10 +1,13 @@
 ---
-license: apache-2.0
-language:
-- en
 base_model:
 - Qwen/Qwen2.5-VL-7B-Instruct
+language:
+- en
 library_name: transformers
+license: apache-2.0
+pipeline_tag: image-to-text
+tags:
+- ocr
 ---
 
 <img alt="olmOCR Logo" src="https://cdn-uploads.huggingface.co/production/uploads/6734d6722769638944a5aa2e/DPsr3ZvRF9v-gdMa4EaHW.png" width="300px" style="margin-left:'auto' margin-right:'auto' display:'block'">
@@ -18,7 +21,7 @@ This is a release of the olmOCR model that's fine tuned from Qwen2.5-VL-7B-Instr
 fine tuned using GRPO RL training to boost its performance at math equations, tables, and other tricky OCR cases.
 
 Quick links:
-- 📃 [Paper](https://olmocr.allenai.org/papers/olmocr.pdf)
+- 📃 [Paper](https://huggingface.co/papers/2510.19817)
 - 🤗 [SFT Dataset](https://huggingface.co/datasets/allenai/olmOCR-mix-1025)
 - 🤗 [RL Dataset](https://huggingface.co/datasets/allenai/olmOCR-synthmix-1025)
 - 🛠️ [Code](https://github.com/allenai/olmocr)
@@ -165,7 +168,16 @@ text_output = processor.tokenizer.batch_decode(
 )
 
 print(text_output)
-# ['---\nprimary_language: en\nis_rotation_valid: True\nrotation_correction: 0\nis_table: False\nis_diagram: False\n---\nolmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models\n\nJake Poz']
+# ['---
+primary_language: en
+is_rotation_valid: True
+rotation_correction: 0
+is_table: False
+is_diagram: False
+---
+olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models
+
+Jake Poz']
 ```
 
 ## License and use