Training BAMM - Bidirectional Autoregressive Motion Model
January 2026
25 min read
Generative Models, Motion Models, Transformers, BAMM
Use arrow keys or click to navigate slides. Press 'F' or Fullscreen icon for best experience.
What You'll Learn
- •Motion Capture data processing and cleaning
- •VQ-VAE architecture for human motion
- •Residual Transformer training
- •Bidirectional Autoregressive modeling explained
- •Evaluating generation quality
Key Concepts Covered
Compressing high-dimensional motion sequences into discrete codebook tokens
Refines motion generation by predicting residual errors
Model that generates sequences considering both past and future context
Resources
Slide Overview
- Data Methodology (Slides 1-8)
- VQ-VAE Implementation (Slides 9-18)
- Transformer Architecture (Slides 19-28)
- Training Pipeline (Slides 29-35)
- Results (Slides 36-40)