metadata
license: bsd-3-clause
tags:
- adafactor
- albef
- batchnorm
- cosine
- kaiming
- matching
- relu
- sliding-window
- small
- tensor-fusion
model_235016012_albef_small.py
Model Overview
A small-scale implementation of the albef architecture, built for matching tasks.
Architecture
- Architecture: albef
- Scale: small
- Attention: sliding window
- Fusion strategy: tensor fusion
- Task head: matching
- Activation: relu
- Normalization: batchnorm
- Initialization: kaiming
Training
- Optimizer: adafactor
- LR scheduler: cosine
Files
model_235016012_albef_small.py— main artifact of this repository
License
See the license field above.