model_665822928_mae_giant.py

Model Overview

A giant-scale implementation of the mae architecture, built for classification tasks.

Architecture

• Architecture: mae
• Scale: giant
• Attention: sparse
• Fusion strategy: tensor fusion
• Task head: classification
• Activation: approx gelu
• Normalization: scalenorm
• Initialization: xavier

Training

• Optimizer: adam
• LR scheduler: linear warmup

Files

• model_665822928_mae_giant.py — main artifact of this repository

License

See the license field above.