model_171418002_blip_large.py

Model Overview

A large-scale implementation of the blip architecture, built for multitask tasks.

Architecture

• Architecture: blip
• Scale: large
• Attention: dilated
• Fusion strategy: co attention
• Task head: multitask
• Activation: swish
• Normalization: groupnorm
• Initialization: xavier uniform

Training

• Optimizer: novograd
• LR scheduler: cosine

Files

• model_171418002_blip_large.py — main artifact of this repository

License

See the license field above.