Add model card and metadata

README.md

@@ -1,3 +1,46 @@
----
-license: mit
----
+---
+license: mit
+pipeline_tag: text-generation
+---
+
+# SHINE: A Scalable In-Context Hypernetwork for Mapping Context to LoRA in a Single Pass
+
+SHINE (Scalable Hyper In-context NEtwork) is a scalable hypernetwork designed to map diverse meaningful contexts into high-quality LoRA adapters for large language models (LLMs) in a single pass. 
+
+- **Paper:** [SHINE: A Scalable In-Context Hypernetwork for Mapping Context to LoRA in a Single Pass](https://huggingface.co/papers/2602.06358)
+- **Repository:** [https://github.com/Yewei-Liu/SHINE](https://github.com/Yewei-Liu/SHINE)
+
+## Description
+
+SHINE overcomes key limitations of prior hypernetworks by reusing the frozen LLM's own parameters in an in-context hypernetwork design. It achieves strong expressive power with a relatively small number of parameters. It transforms in-context knowledge into in-parameter knowledge (LoRA adapters) in a single forward pass, enabling complex question-answering tasks related to a context without requiring the context to be present in the prompt during inference. This approach significantly reduces time, computation, and memory costs compared to standard Supervised Fine-Tuning (SFT).
+
+## Quick Start
+
+### Environment Setup
+
+To set up the environment, follow these steps:
+
+```bash
+conda create -n shine python==3.12 -y
+conda activate shine
+# Change the pytorch version based on your device
+pip install torch==2.5.1 torchvision==0.20.1 torchaudio==2.5.1 --index-url https://download.pytorch.org/whl/cu124
+pip install huggingface==0.0.1 modelscope==1.31.0 transformers==4.57.1 datasets==4.4.1 scikit-learn==1.7.2 hydra-core==1.3.2 tensorboard==2.20.0 openai==2.6.1 rouge==1.0.1 seaborn==0.13.2 matplotlib==3.10.7 multiprocess==0.70.16
+```
+
+### Inference
+
+After downloading the backbone LLM and the hypernetwork checkpoints, you can use the `inference.ipynb` notebook provided in the [official repository](https://github.com/Yewei-Liu/SHINE) for a quick demonstration of the method.
+
+## Citation
+
+If you find this work useful, please cite the paper:
+
+```bibtex
+@article{liu2025shine,
+  title={SHINE: A Scalable In-Context Hypernetwork for Mapping Context to LoRA in a Single Pass},
+  author={Yewei Liu and others},
+  journal={arXiv preprint arXiv:2602.06358},
+  year={2025}
+}
+```