| ---
|
| license: bsd-3-clause
|
| tags:
|
| - batchnorm
|
| - gelu
|
| - generation
|
| - multi-query
|
| - novograd
|
| - onecycle
|
| - orthogonal
|
| - poolformer
|
| - tiny
|
| - tucker
|
| ---
|
|
|
| # model_416623696_poolformer_tiny.py
|
|
|
| ## Model Overview
|
|
|
| A **tiny**-scale implementation of the **poolformer** architecture, built for **generation** tasks.
|
|
|
| ## Architecture
|
|
|
| - **Architecture**: poolformer
|
| - **Scale**: tiny
|
| - **Attention**: multi query
|
| - **Fusion strategy**: tucker
|
| - **Task head**: generation
|
| - **Activation**: gelu
|
| - **Normalization**: batchnorm
|
| - **Initialization**: orthogonal
|
|
|
| ## Training
|
|
|
| - **Optimizer**: novograd
|
| - **LR scheduler**: onecycle
|
|
|
| ## Files
|
|
|
| - `model_416623696_poolformer_tiny.py` — main artifact of this repository
|
|
|
| ## License
|
|
|
| See the license field above.
|
|
|
