model_284810674_coca_xlarge.py

Model Overview

A xlarge-scale implementation of the coca architecture, built for matching tasks.

Architecture

• Architecture: coca
• Scale: xlarge
• Attention: sparse
• Fusion strategy: tucker
• Task head: matching
• Activation: approx gelu
• Normalization: instancenorm
• Initialization: trunc normal

Training

• Optimizer: rmsprop
• LR scheduler: linear warmup

Files

• model_284810674_coca_xlarge.py — main artifact of this repository

License

See the license field above.