model.py
Model Overview
A large-scale implementation of the swin t architecture, built for generation tasks.
Architecture
- Architecture: swin t
- Scale: large
- Attention: multi query
- Fusion strategy: co attention
- Task head: generation
- Activation: approx gelu
- Normalization: groupnorm
- Initialization: trunc normal
Training
- Optimizer: adafactor
- LR scheduler: constant warmup
Files
model.py — main artifact of this repository
License
See the license field above.
