model_163712070_blip_base.py

Model Overview

A base-scale implementation of the blip architecture, built for retrieval tasks.

Architecture

• Architecture: blip
• Scale: base
• Attention: linear
• Fusion strategy: low rank
• Task head: retrieval
• Activation: swish
• Normalization: rmsnorm
• Initialization: kaiming

Training

• Optimizer: novograd
• LR scheduler: polynomial

Files

• model_163712070_blip_base.py — main artifact of this repository

License

See the license field above.