Skip to content

fastc

Lightweight text classification on CPU with embeddings

Visit official site

What fastc is

The tool is described as a simple, lightweight text classifier built on large language model embeddings, focused on CPU execution and cosine similarity rather than fine-tuning.

What you can do with it

  • Generate embeddings with efficient models on CPU
  • Classify text by cosine similarity
  • Share one model embedding across classifiers
  • Load pre-trained models from a folder or a hub

Getting started

  • Install the tool through the package manager
  • Prepare text data for training a classifier
  • Train a classifier and save the model

Pros & cons

✓ What we like

  • Several classifiers run without extra overhead
  • Models can be trained, exported and published to Hugging Face

! What to watch out for

  • Classification relies on cosine similarity
  • The named execution target is CPU

FAQ

Is fine-tuning needed?

The material states cosine similarity is used instead of fine-tuning.

Where can models be published?

The feature list names the Hugging Face model hub.

Last reviewed: 2026-11-15

More AI office assistant tools

View all →

How we review