Home

FasterAI is an open-source library for neural network compression, offering efficient implementations of various compression techniques.

Compression techniques in a single line of code

from fasterai.sparse.all import *
from fastai.vision.all import *

learn = Learner(dls, model)
sp_cb = SparsifyCallback(sparsity, granularity, context, criteria, schedule)

learn.fit(epochs, lr, cbs=sp_cb)

Sparsification

Pruning

Distillation

Regularization

Quantization

Home

Welcome to FasterAI

Compression techniques in a single line of code

Make your models smaller, faster, cheaper and greener

Supported methods