| ---
|
| license: apache-2.0
|
| tags:
|
| - adafactor
|
| - classification
|
| - cosine
|
| - flamingo
|
| - flash
|
| - kaiming
|
| - large
|
| - mish
|
| - scalenorm
|
| - tucker
|
| ---
|
|
|
| # inference.py
|
|
|
| ## Model Overview
|
|
|
| A **large**-scale implementation of the **flamingo** architecture, built for **classification** tasks.
|
|
|
| ## Architecture
|
|
|
| - **Architecture**: flamingo
|
| - **Scale**: large
|
| - **Attention**: flash
|
| - **Fusion strategy**: tucker
|
| - **Task head**: classification
|
| - **Activation**: mish
|
| - **Normalization**: scalenorm
|
| - **Initialization**: kaiming
|
|
|
| ## Training
|
|
|
| - **Optimizer**: adafactor
|
| - **LR scheduler**: cosine
|
|
|
| ## Files
|
|
|
| - `inference.py` — main artifact of this repository
|
|
|
| ## License
|
|
|
| See the license field above.
|
|
|
