Miau Labs / Insight

DiScoFormer: One Transformer for Density and Score, Across Distributions

DiScoFormer is a transformer model that estimates density and score of distributions in a single forward pass without retraining. It outperforms kernel density estimation (KDE) in both density and score estimation, especially in high-dimensional spaces.

DiScoFormer is a transformer model that estimates density and score of distributions in a single forward pass without retraining. It outperforms kernel density estimation (KDE) in both density and score estimation, especially in high-dimensional spaces. DiScoFormer is a transformer model that offers an effective solution for density and score estimation of distributions in high-dimensional spaces.

DiScoFormer is a transformer model that estimates density and score of distributions in a single forward pass without retraining. It outperforms kernel density estimation (KDE) in both density and score estimation, especially in high-dimensional spaces.

  • DiScoFormer is a transformer model that estimates density and score of distributions in a single forward pass without retraining.
  • It outperforms kernel density estimation (KDE) in both density and score estimation, especially in high-dimensional spaces.
  • DiScoFormer uses a shared backbone with two output heads, one for density and one for score, leveraging their mathematical relationship.
Miau Labs takeDiScoFormer is a transformer model that offers an effective solution for density and score estimation of distributions in high-dimensional spaces.