metadata
license: bsd-3-clause
tags:
- clip
- constant-warmup
- large
- layernorm
- linear
- matching
- relu
- sgd
- tensor-fusion
- trunc-normal
main.py
Model Overview
A large-scale implementation of the clip architecture, built for matching tasks.
Architecture
- Architecture: clip
- Scale: large
- Attention: linear
- Fusion strategy: tensor fusion
- Task head: matching
- Activation: relu
- Normalization: layernorm
- Initialization: trunc normal
Training
- Optimizer: sgd
- LR scheduler: constant warmup
Files
main.py— main artifact of this repository
License
See the license field above.