model_466644827_swin_t_large.py

Model Overview

A large-scale implementation of the swin t architecture, built for retrieval tasks.

Architecture

• Architecture: swin t
• Scale: large
• Attention: flash
• Fusion strategy: bilinear
• Task head: retrieval
• Activation: swish
• Normalization: scalenorm
• Initialization: orthogonal

Training

• Optimizer: sgd
• LR scheduler: constant warmup

Files

• model_466644827_swin_t_large.py — main artifact of this repository

License

See the license field above.