model_643734008_mae_small.py

Model Overview

A small-scale implementation of the mae architecture, built for retrieval tasks.

Architecture

• Architecture: mae
• Scale: small
• Attention: sliding window
• Fusion strategy: bilinear
• Task head: retrieval
• Activation: gelu tanh
• Normalization: layernorm
• Initialization: orthogonal

Training

• Optimizer: adam
• LR scheduler: linear warmup

Files

• model_643734008_mae_small.py — main artifact of this repository

License

See the license field above.