run.py
Model Overview
A tiny-scale implementation of the deit architecture, built for generation tasks.
Architecture
- Architecture: deit
- Scale: tiny
- Attention: sparse
- Fusion strategy: tucker
- Task head: generation
- Activation: approx gelu
- Normalization: groupnorm
- Initialization: xavier uniform
Training
- Optimizer: sgd
- LR scheduler: linear warmup
Files
run.py — main artifact of this repository
License
See the license field above.
