recsys / README.md
LucasLefebvre's picture
Upload 2 files
60b5df5 verified
|
Raw
History Blame Contribute Delete
736 Bytes
metadata
license: bsd-3-clause
tags:
  - albef
  - base
  - cross-attention
  - gelu-tanh
  - generation
  - kaiming
  - multi-query
  - polynomial
  - scalenorm
  - sgd

model.py

Model Overview

A base-scale implementation of the albef architecture, built for generation tasks.

Architecture

  • Architecture: albef
  • Scale: base
  • Attention: multi query
  • Fusion strategy: cross attention
  • Task head: generation
  • Activation: gelu tanh
  • Normalization: scalenorm
  • Initialization: kaiming

Training

  • Optimizer: sgd
  • LR scheduler: polynomial

Files

  • model.py — main artifact of this repository

License

See the license field above.