model_076617287_perceiver_nano.py

Model Overview

A nano-scale implementation of the perceiver architecture, built for retrieval tasks.

Architecture

• Architecture: perceiver
• Scale: nano
• Attention: flash
• Fusion strategy: low rank
• Task head: retrieval
• Activation: gelu
• Normalization: groupnorm
• Initialization: orthogonal

Training

• Optimizer: novograd
• LR scheduler: constant warmup

Files

• model_076617287_perceiver_nano.py — main artifact of this repository

License

See the license field above.