Nepandey's picture
Upload 2 files
74cf6a6 verified
|
Raw History Blame Contribute Delete
851 Bytes
metadata
license: bsd-3-clause
tags:
  - adam
  - base
  - classification
  - gelu
  - grouped-query
  - groupnorm
  - kaiming-normal
  - polynomial
  - tensor-fusion
  - tiny-transformer

model_315126851_tiny_transformer_base.py

Model Overview

A base-scale implementation of the tiny transformer architecture, built for classification tasks.

Architecture

  • Architecture: tiny transformer
  • Scale: base
  • Attention: grouped query
  • Fusion strategy: tensor fusion
  • Task head: classification
  • Activation: gelu
  • Normalization: groupnorm
  • Initialization: kaiming normal

Training

  • Optimizer: adam
  • LR scheduler: polynomial

Files

  • model_315126851_tiny_transformer_base.py — main artifact of this repository

License

See the license field above.