model_220227662_clip_giant.py

Model Overview

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

Architecture

• Architecture: clip
• Scale: giant
• Attention: flash
• Fusion strategy: concat mlp
• Task head: contrastive
• Activation: mish
• Normalization: batchnorm
• Initialization: kaiming normal

Training

• Optimizer: novograd
• LR scheduler: polynomial

Files

• model_220227662_clip_giant.py — main artifact of this repository

License

See the license field above.