Add metadata and improve model card (#1)

- Add metadata and improve model card (3631d3332a9d9ebbf6937554909c5d7b0ed58602)


Co-authored-by: Niels Rogge <nielsr@users.noreply.huggingface.co>

README.md

@@ -1,58 +1,34 @@
-# Skip-BART
-
-The description is generated by Grok3.
-
-## Model Details
-
-- **Model Name**: Skip-BART
-
-- **Model Type**: Transformer-based model (BART architecture) for automatic stage lighting control
-
-- **Version**: 1.0
-
-- **Release Date**: August 2025
-
-- **Developers**: Zijian Zhao, Dian Jin
-
-- **Organization**: HKUST, PolyU
-
-- **License**: Apache License 2.0
-
-- **Paper**: [Automatic Stage Lighting Control: Is it a Rule-Driven Process or Generative Task?](https://arxiv.org/abs/2506.01482)
+---
+license: apache-2.0
+pipeline_tag: other
+datasets:
+- RS2002/RPMC-L2
+tags:
+- stage-lighting
+- generative-task
+- music-to-light
+---
 
-- **Citation:**
-
-  ```
-  @article{zhao2025automatic,
-    title={Automatic Stage Lighting Control: Is it a Rule-Driven Process or Generative Task?},
-    author={Zhao, Zijian and Jin, Dian and Zhou, Zijing and Zhang, Xiaoyu},
-    journal={arXiv preprint arXiv:2506.01482},
-    year={2025}
-  }
-  ```
+# Skip-BART
 
-- **Contact**: zzhaock@connect.ust.hk
+Skip-BART is an end-to-end generative model designed for **Automatic Stage Lighting Control (ASLC)**. Unlike traditional rule-based methods, Skip-BART conceptualizes lighting control as a generative task, learning directly from professional lighting engineers to predict vivid, human-like lighting sequences synchronized with music.
 
-- **Repository**: https://github.com/RS2002/Skip-BART
+This model was presented in the paper [Automatic Stage Lighting Control: Is it a Rule-Driven Process or Generative Task?](https://huggingface.co/papers/2506.01482).
 
-## Model Description
+- **Repository**: [https://github.com/RS2002/Skip-BART](https://github.com/RS2002/Skip-BART)
+- **Dataset**: [RS2002/RPMC-L2](https://huggingface.co/datasets/RS2002/RPMC-L2)
 
-Skip-BART is a transformer-based model built on the Bidirectional and Auto-Regressive Transformers (BART) architecture, designed for automatic stage lighting control. It generates lighting sequences synchronized with music input, treating stage lighting as a generative task. The model processes music data in an octuple format and outputs lighting control parameters, leveraging a skip-connection-enhanced BART structure for improved performance.
+## Model Details
 
-- **Architecture**: BART with skip connections
-- **Input Format**: Encoder input (batch_size, length, 512), decoder input (batch_size, length, 2), attention masks (batch_size, length)
-- **Output Format**: Hidden states of dimension [batch_size, length, 1024]
-- **Hidden Size**: 1024
-- **Training Objective**: Pre-training on music data, followed by fine-tuning for lighting sequence generation
-- **Tasks Supported**: Stage lighting sequence generation
+- **Model Type**: Transformer-based model (BART architecture) with skip connections.
+- **Task**: Stage lighting sequence generation (predicting light hue and intensity).
+- **Architecture**: BART-based structure enhanced with a novel skip-connection mechanism to strengthen the relationship between musical frames and lighting states.
+- **Input Format**: Encoder input (batch_size, length, 512) for audio features; Decoder input (batch_size, length, 2) for lighting parameters.
+- **Output Format**: Hidden states representing lighting control parameters (dimension 1024).
 
 ## Training Data
 
-The model was trained on the **RPMC-L2** dataset:
-
-- **Dataset Source**: [RPMC-L2](https://zenodo.org/records/14854217?token=eyJhbGciOiJIUzUxMiJ9.eyJpZCI6IjM5MDcwY2E5LTY0MzUtNGZhZC04NzA4LTczMjNhNTZiOGZmYSIsImRhdGEiOnt9LCJyYW5kb20iOiI1YWRkZmNiMmYyOGNiYzI4ZWUxY2QwNTAyY2YxNTY4ZiJ9.0Jr6GYfyyn02F96eVpkjOtcE-MM1wt-_ctOshdNGMUyUKI15-9Rfp9VF30_hYOTqv_9lLj-7Wj0qGyR3p9cA5w)
-- **Description**: Contains music and corresponding stage lighting data in a format suitable for training Skip-BART.
-- **Details**: Refer to the [paper](https://arxiv.org/abs/2506.01482) for dataset specifics.
+The model was trained on the **RPMC-L2** dataset, a self-collected dataset containing music and corresponding stage lighting data synchronized within a frame grid.
 
 ## Usage
 
@@ -64,6 +40,8 @@ git clone https://huggingface.co/RS2002/Skip-BART
 
 ### Example Code
 
+The following snippet demonstrates how to load the model and perform a forward pass (requires `model.py` from the official repository).
+
 ```python
 import torch
 from model import Skip_BART
@@ -80,4 +58,19 @@ decoder_attention_mask = torch.zeros((2, 1024))
 # Forward pass
 output = model(x_encoder, x_decoder, encoder_attention_mask, decoder_attention_mask)
 print(output.size())  # Output: [2, 1024, 1024]
-```
+```
+
+## Citation
+
+```bibtex
+@article{zhao2025automatic,
+  title={Automatic Stage Lighting Control: Is it a Rule-Driven Process or Generative Task?},
+  author={Zhao, Zijian and Jin, Dian and Zhou, Zijing and Zhang, Xiaoyu},
+  journal={arXiv preprint arXiv:2506.01482},
+  year={2025}
+}
+```
+
+## Contact
+
+Zijian Zhao: zzhaock@connect.ust.hk