metadata
license: apache-2.0
tags:
- adafactor
- cosine
- gated-fusion
- gelu-tanh
- grouped-query
- poolformer
- retrieval
- rmsnorm
- small
- xavier-uniform
model.py
Model Overview
A small-scale implementation of the poolformer architecture, built for retrieval tasks.
Architecture
- Architecture: poolformer
- Scale: small
- Attention: grouped query
- Fusion strategy: gated fusion
- Task head: retrieval
- Activation: gelu tanh
- Normalization: rmsnorm
- Initialization: xavier uniform
Training
- Optimizer: adafactor
- LR scheduler: cosine
Files
model.py— main artifact of this repository
License
See the license field above.