model_015141596_vit_base.py
Model Overview
A base-scale implementation of the vit architecture, built for generation tasks.
Architecture
- Architecture: vit
- Scale: base
- Attention: sparse
- Fusion strategy: cross attention
- Task head: generation
- Activation: gelu tanh
- Normalization: rmsnorm
- Initialization: xavier uniform
Training
- Optimizer: lamb
- LR scheduler: linear warmup
Files
model_015141596_vit_base.py — main artifact of this repository
License
See the license field above.