metadata
license: apache-2.0
tags:
- batchnorm
- contrastive
- cross-attention
- flamingo
- grouped-query
- lion
- onecycle
- swish
- xavier-uniform
- xlarge
main.py
Model Overview
A xlarge-scale implementation of the flamingo architecture, built for contrastive tasks.
Architecture
- Architecture: flamingo
- Scale: xlarge
- Attention: grouped query
- Fusion strategy: cross attention
- Task head: contrastive
- Activation: swish
- Normalization: batchnorm
- Initialization: xavier uniform
Training
- Optimizer: lion
- LR scheduler: onecycle
Files
main.py— main artifact of this repository
License
See the license field above.