metadata
license: mit
tags:
- bilinear
- flamingo
- gelu
- generation
- kaiming-normal
- linear
- novograd
- polynomial
- scalenorm
- small
model_121748246_flamingo_small.py
Model Overview
A small-scale implementation of the flamingo architecture, built for generation tasks.
Architecture
- Architecture: flamingo
- Scale: small
- Attention: linear
- Fusion strategy: bilinear
- Task head: generation
- Activation: gelu
- Normalization: scalenorm
- Initialization: kaiming normal
Training
- Optimizer: novograd
- LR scheduler: polynomial
Files
model_121748246_flamingo_small.py— main artifact of this repository
License
See the license field above.