model_424437810_poolformer_base.py

Model Overview

A base-scale implementation of the poolformer architecture, built for matching tasks.

Architecture

• Architecture: poolformer
• Scale: base
• Attention: flash
• Fusion strategy: cross attention
• Task head: matching
• Activation: gelu
• Normalization: instancenorm
• Initialization: orthogonal

Training

• Optimizer: adafactor
• LR scheduler: onecycle

Files

• model_424437810_poolformer_base.py — main artifact of this repository

License

See the license field above.