train.py
Model Overview
A huge-scale implementation of the mixer architecture, built for generation tasks.
Architecture
- Architecture: mixer
- Scale: huge
- Attention: sliding window
- Fusion strategy: tensor fusion
- Task head: generation
- Activation: gelu tanh
- Normalization: batchnorm
- Initialization: kaiming
Training
- Optimizer: adamw
- LR scheduler: constant warmup
Files
train.py— main artifact of this repository
License
See the license field above.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support