model_628448749_beit_giant.py

Model Overview

A giant-scale implementation of the beit architecture, built for matching tasks.

Architecture

• Architecture: beit
• Scale: giant
• Attention: sparse
• Fusion strategy: gated fusion
• Task head: matching
• Activation: gelu
• Normalization: rmsnorm
• Initialization: orthogonal

Training

• Optimizer: adamw
• LR scheduler: constant warmup

Files

• model_628448749_beit_giant.py — main artifact of this repository

License

See the license field above.