Skip to content

Faded-Nebula/FMClassifier

Repository files navigation

Flow Matching Zero-Shot Classifier

Requirements

You can install the environment by:

conda env create -f env.yaml

Also install our version of diffusers:

cd diffusers
pip install -e .

Usage

  1. Use train.py to train a model on MNIST now.
  2. Use eval.py to view the generation examples.
  3. Use classifier.py to test the classification performance.
  4. Use flux.py to do zero-shot classification.

About

Implementation for Your Flow Matching Model is Implicitly A Zero-Shot Classifier

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors