Add pipeline_tag and library_name, remove duplicate base model entry, fix typo

This PR adds the `pipeline_tag` (set to `text-generation`) and `library_name` (set to `vllm`) to the model card metadata. This PR also:
- fixes a typo
- adds a link to the paper to the model card
- removes a duplicated base model entry

README.md

@@ -1,22 +1,25 @@
 ---
-license: apache-2.0
 base_model:
 - Satori-reasoning/Satori-SWE-SFT-32B
+license: apache-2.0
+pipeline_tag: text-generation
+library_name: vllm
+datasets:
+- Satori-reasoning/Satori-SWE-two-stage-SFT-data
+- Satori-reasoning/Satori-SWE-RL-data
 ---
 
 # Satori‑SWE‑RL‑32B
 
 ## Overview
 
-🚀 **Satori-SWE-RL-32B** is trained specifically to resolve software engineering tasks efficiently, using our proposed [**EvoScale**](https://arxiv.org/pdf/2505.23604) test-time scaling technique, and a novel training framework: two-stage SFT and RL. The model can iteratively self-improve its own generation to progressively write a better patch.
-
+🚀 **Satori-SWE-RL-32B** is trained specifically to resolve software engineering tasks efficiently, using our proposed [**EvoScale**](https://arxiv.org/abs/2505.23604) test-time scaling technique, and a novel training framework: two-stage SFT and RL. The model can iteratively self-improve its own generation to progressively write a better patch.
 
 ## Training Data
 
 - **SFT Dataset**: [Satori SFT Dataset](https://huggingface.co/datasets/Satori-reasoning/Satori-SWE-two-stage-SFT-data)
 - **RL Dataset**: [Satori RL Dataset](https://huggingface.co/datasets/Satori-reasoning/Satori-SWE-RL-data)
 
-
 ## Resources
 
 🔗 **GitHub Repository**: [Satori-SWE](https://github.com/Satori-Reasoning/Satori-SWE)
@@ -25,7 +28,6 @@ base_model:
 
 🔗 **Research Paper**: [Paper](https://arxiv.org/abs/2505.23604)
 
-
 ## Prompt Template
 ````python
 classical_sft_prompt = """You are an expert software engineer and seasoned code reviewer, specializing in bug localization and code optimization within real-world code repositories. Your strengths lie in understanding complex codebase structures and precisely identifying and modifying the relevant parts of the code to resolve issues. You also excel at articulating your reasoning process in a coherent, step-by-step manner that leads to efficient and correct bug fixes.
@@ -249,8 +251,6 @@ Please provide your response below.
 """
 ````
 
-
-
 ## Usage: Toy Example
 
 ````python
@@ -445,8 +445,6 @@ for mutation_completion in mutation_completions:
 
 ````
 
-
-
 ## Citation
 
 If you find this model useful, please cite our paper:
@@ -461,4 +459,4 @@ If you find this model useful, please cite our paper:
       primaryClass={cs.CL},
       url={https://arxiv.org/abs/2505.23604}, 
 }
-```
+```