Multiplierless MLP Using Successive Vector Approximation in Post-Training Quantization

Run experiments using the main.py file. You'll need Python >= 3.10, numpy, pytorch, numba, and matplolib.

You must first train a model using:

$ python main.py -m train

Depending on the machine and whether you use GPU, this should take a few hours.

Fig. 2 can be reproduced with the following:

$ python main.py -m run

For Fig. 3, run the following command:

$ python main.py -m run2

For Fig. 4, you must have the quantized models from -m run and run the command:

$ python main.py -m run3