model_072318702_cnn_transformer_giant.py

Model Overview

A giant-scale implementation of the cnn transformer architecture, built for contrastive tasks.

Architecture

• Architecture: cnn transformer
• Scale: giant
• Attention: sliding window
• Fusion strategy: tucker
• Task head: contrastive
• Activation: gelu tanh
• Normalization: rmsnorm
• Initialization: kaiming

Training

• Optimizer: lion
• LR scheduler: linear warmup

Files

• model_072318702_cnn_transformer_giant.py — main artifact of this repository

License

See the license field above.