dizolivemint/motion-encoder-decoder

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

🧠 Motion Encoder Decoder ML Pipeline

An interactive Gradio-based machine learning pipeline for generating, training, and testing encoder-decoder models on simulated physics trajectories.

This PyTorch-based system models motion dynamics such as projectile paths and bouncing objects using a sequence-to-sequence architecture.

Features

• —⚙️ Dataset generation (custom physics simulations)
• —🧪 Training with optional early stopping
• —📈 Input sensitivity testing
• —🔮 Real-time predictions and trajectory visualizations
• —📤 Upload / 📥 Download of models and datasets (in /tmp)

Try It Out

1. 1.Select a physics type
2. 2.Generate or upload a dataset
3. 3.Train a model or upload a pretrained .pth
4. 4.Visualize predictions from dynamic input sliders

Built With

• —Python 3.13
• —PyTorch
• —Gradio
• —Matplotlib
• —NumPy

👨‍💻 Developed by Miles Exner