Update usage instructions with huggingface_hub download and multiple usage options

README.md

@@ -74,22 +74,66 @@ This model separates ECG signals into class-invariant **content** (beat morpholo
 
 ## Usage
 
+### Option 1: Interactive Demo (Easiest)
+
+Try the model directly in your browser — no code needed:
+
+👉 **[Launch Demo](https://huggingface.co/spaces/TharakaDil2001/ecg-augmentation-demo)**
+
+Upload an ECG (`.npy` or `.csv`, 2500 samples at 250 Hz) or browse pre-loaded examples.
+
+### Option 2: Download & Use in Python
+
 ```python
+from huggingface_hub import hf_hub_download
 import torch
 
+# Download model files from Hugging Face
+diffusion_path = hf_hub_download(
+    repo_id="TharakaDil2001/diffusion-ecg-augmentation",
+    filename="diffusion_model.pth"
+)
+classifier_path = hf_hub_download(
+    repo_id="TharakaDil2001/diffusion-ecg-augmentation",
+    filename="classifier_model.pth"
+)
+
 # Load the diffusion model checkpoint
-checkpoint = torch.load("diffusion_model.pth", map_location="cpu")
+checkpoint = torch.load(diffusion_path, map_location="cpu")
 
 # The checkpoint contains:
-# - content_encoder state dict
-# - style_encoder state dict  
-# - unet state dict
-# - config object with all hyperparameters
+# - checkpoint['content_encoder'] → Content Encoder state dict
+# - checkpoint['style_encoder']   → Style Encoder state dict
+# - checkpoint['unet']            → UNet state dict
+# - checkpoint['config']          → Training config with all hyperparameters
+
+# Load the classifier checkpoint
+cls_checkpoint = torch.load(classifier_path, map_location="cpu")
+# - cls_checkpoint['model_state_dict'] → AFibResLSTM state dict
+
+# To use the full pipeline, clone the repository:
+# git clone https://github.com/vlbthambawita/PERA_AF_Detection.git
+# See: diffusion_pipeline/final_pipeline/ for model architectures
+```
+
+### Option 3: Clone the Full Pipeline
 
-# See the full pipeline code at:
-# https://github.com/vlbthambawita/PERA_AF_Detection/tree/main/diffusion_pipeline/final_pipeline
+```bash
+# Clone the full codebase with all model architectures
+git clone https://github.com/vlbthambawita/PERA_AF_Detection.git
+cd PERA_AF_Detection/diffusion_pipeline/final_pipeline/
+
+# Download weights
+pip install huggingface_hub
+python -c "
+from huggingface_hub import hf_hub_download
+hf_hub_download('TharakaDil2001/diffusion-ecg-augmentation', 'diffusion_model.pth', local_dir='.')
+hf_hub_download('TharakaDil2001/diffusion-ecg-augmentation', 'classifier_model.pth', local_dir='.')
+"
 ```
 
+> **Note**: The model architectures (DiffStyleTS, AFibResLSTM) are defined in the repository code. You need the architecture classes to instantiate the models before loading the state dicts.
+
 ## Citation
 
 ```bibtex