Vibeleaderboard
Index / tool
Visit github.com
Category
AI Tools
Rank
No. 3617Tools index

Previous survey · No. 3411 ·

Type
TOOL
Use case
Models: Train & Run · Data Processing
GitHub
128 stars
Date

About

Jeffy is a collection of small classifiers that run locally on a regular computer without a GPU. It ships with 13 pretrained classifiers covering spam detection, intent routing, topic classification and sentiment analysis, and one that plays Doom. You can also train your own classifiers with it.

At a glance

Jeffy provides pretrained classifiers for text and structured inputs that run on a CPU and can be retrained. It is worth trying if you have fixed labels and want an inexpensive baseline before using a general-purpose model.

Good fit

Developers building intent routing or other repeated classification tasks.

Consider something else when

Use a different approach if the output must be an open-ended answer or none of the available label sets fits.

Start here

Pick a classifier and test it on a small labeled sample, including inputs outside its expected categories. Inspect errors before routing live requests based on its predictions.

Cost and requirements

MIT-licensed project. The documented setup uses a Python package and CPU inference. Check the applicable terms for the classifier or data you choose.

Limits to know

Pretrained results depend on the label set and training data. Check performance on your language and domain before relying on a prediction.

Tags

classifiersmllocalno-gpusentiment-analysisspam-detection

Tech Stack

Python

Comments (0)

No comments yet

Editorially curated, with community endorsements as a secondary signal. Corrections welcome.