model_627480026_beit_giant.py

Model Overview

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

Architecture

• Architecture: beit
• Scale: giant
• Attention: standard
• Fusion strategy: bilinear
• Task head: multitask
• Activation: gelu
• Normalization: layernorm
• Initialization: trunc normal

Training

• Optimizer: lion
• LR scheduler: constant warmup

Files

• model_627480026_beit_giant.py — main artifact of this repository

License

See the license field above.