Skip to main content
Sobolsoft software catalog

Web App

NEW

Pack Token Sequences Into Training Blocks

Build fixed-length token blocks with explicit masks.

Web Application
Try Webapp

Screenshot

Inspect the interface

Description

What is this software?

Build fixed-length token blocks with explicit masks.

Features

  • Pack local token IDs or locally tokenized text into fixed-length training blocks.
  • Control document boundaries, BOS/EOS insertion, padding, attention masks and masked labels, with source spans.

Build fixed-length token blocks with explicit masks.

Features

  • Pack local token IDs or locally tokenized text into fixed-length training blocks.
  • Control document boundaries, BOS/EOS insertion, padding, attention masks and masked labels, with source spans.