Improve model card with detailed description, sample usage, citation, and acknowledgements

README.md

@@ -12,9 +12,47 @@ tags:
 
 # Model Card for Lite-Whisper large-v3-fast
 
-<!-- Provide a quick summary of what the model is/does. -->
+LiteASR introduces Lite-Whisper, a low-rank compression scheme for ASR encoders, significantly reducing inference costs while maintaining transcription accuracy. This approach leverages strong low-rank properties observed in intermediate activations of deep encoder-decoder architectures, particularly OpenAI's Whisper. By applying principal component analysis (PCA) with a small calibration dataset, LiteASR approximates linear transformations with a chain of low-rank matrix multiplications, and further optimizes self-attention to work in reduced dimensionality. Evaluation results show that our method can compress Whisper large-v3's encoder size by over 50%, matching Whisper medium's size with better transcription accuracy, thereby establishing a new Pareto frontier of accuracy and efficiency.
 
-Lite-Whisper is a compressed version of OpenAI Whisper with LiteASR. See our [GitHub repository](https://github.com/efeslab/LiteASR) and [paper](https://arxiv.org/abs/2502.20583) for details. The paper is also available on Hugging Face: [Link to Hugging Face Paper Page](https://hf.co/papers/2502.20583)
+Lite-Whisper is a compressed version of OpenAI Whisper with LiteASR. For technical details and full methodology, refer to our [paper](https://arxiv.org/abs/2502.20583) (also available on Hugging Face: [Link to Hugging Face Paper Page](https://hf.co/papers/2502.20583)) and the [LiteASR GitHub repository](https://github.com/efeslab/LiteASR).
+
+## Quick Start
+The easiest way to run our model is to use our integration with HuggingFace Transformers library.
+We provide model weights for the compressed version of OpenAI Whisper series [here](https://huggingface.co/efficient-speech).
+
+```python
+import librosa 
+import torch
+from transformers import AutoProcessor, AutoModel
+
+device = "cuda:0"
+dtype = torch.float16
+
+# load the compressed Whisper model
+model = AutoModel.from_pretrained(
+    "efficient-speech/lite-whisper-large-v3-turbo", 
+    trust_remote_code=True, 
+)
+model.to(dtype).to(device)
+
+# we use the same processor as the original model
+processor = AutoProcessor.from_pretrained("openai/whisper-large-v3")
+
+# set the path to your audio file
+path = "path/to/audio.wav"
+audio, _ = librosa.load(path, sr=16000)
+
+input_features = processor(audio, sampling_rate=16000, return_tensors="pt").input_features
+input_features = input_features.to(dtype).to(device)
+
+predicted_ids = model.generate(input_features)
+transcription = processor.batch_decode(
+    predicted_ids, 
+    skip_special_tokens=True
+)[0]
+
+print(transcription)
+```
 
 ## Benchmark Results
 
@@ -32,4 +70,26 @@ Following is the average word error rate (WER) evaluated on the [ESB datasets](h
 | [lite-whisper-large-v3-turbo](https://huggingface.co/efficient-speech/lite-whisper-large-v3-turbo) | 12.6 | 374M | 172M |
 | [lite-whisper-large-v3-turbo-fast](https://huggingface.co/efficient-speech/lite-whisper-large-v3-turbo-fast) | 20.1 | 313M | 172M |
 | &nbsp; | &nbsp; | &nbsp; | &nbsp; |
-| [whisper-medium](https://huggingface.co/openai/whisper-medium) | 14.8 | 306M | 457M |
+| [whisper-medium](https://huggingface.co/openai/whisper-medium) | 14.8 | 306M | 457M |
+
+## Acknowledgement
+
+- [OpenAI Whisper](https://github.com/openai/whisper)
+- [MLX Whisper](https://github.com/ml-explore/mlx-examples/tree/main/whisper)
+- [ASR Leaderboard](https://github.com/huggingface/open_asr_leaderboard)
+
+## Citation
+
+If you use LiteASR in your research, please cite the following paper:
+
+```
+@misc{kamahori2025liteasrefficientautomaticspeech,
+      title={LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation}, 
+      author={Keisuke Kamahori and Jungo Kasai and Noriyuki Kojima and Baris Kasikci},
+      year={2025},
+      eprint={2502.20583},
+      archivePrefix={arXiv},
+      primaryClass={cs.LG},
+      url={https://arxiv.org/abs/2502.20583}, 
+}
+```