roberta-classifier6 / README.md
schmid-tlaura's picture
Upload 2 files
205d112 verified
|
Raw History Blame Contribute Delete
755 Bytes
metadata
license: apache-2.0
tags:
  - adamw
  - approx-gelu
  - base
  - constant-warmup
  - mobilevit
  - retrieval
  - rmsnorm
  - sliding-window
  - tensor-fusion
  - xavier

eval.py

Model Overview

A base-scale implementation of the mobilevit architecture, built for retrieval tasks.

Architecture

  • Architecture: mobilevit
  • Scale: base
  • Attention: sliding window
  • Fusion strategy: tensor fusion
  • Task head: retrieval
  • Activation: approx gelu
  • Normalization: rmsnorm
  • Initialization: xavier

Training

  • Optimizer: adamw
  • LR scheduler: constant warmup

Files

  • eval.py — main artifact of this repository

License

See the license field above.