| ---
|
| license: apache-2.0
|
| tags:
|
| - adafactor
|
| - classification
|
| - exponential
|
| - instancenorm
|
| - relu
|
| - sliding-window
|
| - swin-t
|
| - tensor-fusion
|
| - trunc-normal
|
| - xlarge
|
| ---
|
|
|
| # main.py
|
|
|
| ## Model Overview
|
|
|
| A **xlarge**-scale implementation of the **swin t** architecture, built for **classification** tasks.
|
|
|
| ## Architecture
|
|
|
| - **Architecture**: swin t
|
| - **Scale**: xlarge
|
| - **Attention**: sliding window
|
| - **Fusion strategy**: tensor fusion
|
| - **Task head**: classification
|
| - **Activation**: relu
|
| - **Normalization**: instancenorm
|
| - **Initialization**: trunc normal
|
|
|
| ## Training
|
|
|
| - **Optimizer**: adafactor
|
| - **LR scheduler**: exponential
|
|
|
| ## Files
|
|
|
| - `main.py` — main artifact of this repository
|
|
|
| ## License
|
|
|
| See the license field above.
|
|
|
