| ---
|
| license: cc-by-4.0
|
| tags:
|
| - approx-gelu
|
| - classification
|
| - cosine
|
| - giant
|
| - instancenorm
|
| - novograd
|
| - sparse
|
| - tensor-fusion
|
| - vit
|
| - xavier-uniform
|
| ---
|
|
|
| # pipeline.py
|
|
|
| ## Model Overview
|
|
|
| A **giant**-scale implementation of the **vit** architecture, built for **classification** tasks.
|
|
|
| ## Architecture
|
|
|
| - **Architecture**: vit
|
| - **Scale**: giant
|
| - **Attention**: sparse
|
| - **Fusion strategy**: tensor fusion
|
| - **Task head**: classification
|
| - **Activation**: approx gelu
|
| - **Normalization**: instancenorm
|
| - **Initialization**: xavier uniform
|
|
|
| ## Training
|
|
|
| - **Optimizer**: novograd
|
| - **LR scheduler**: cosine
|
|
|
| ## Files
|
|
|
| - `pipeline.py` — main artifact of this repository
|
|
|
| ## License
|
|
|
| See the license field above.
|
|
|
