import logging
import numpy as np
import matplotlib.pyplot as plt
import pdb
from dorefa_ops import get_dorefa
from math_ops import *
from random import randint
logging.getLogger().setLevel(logging.DEBUG)
BITW = 1
BITA = 1
BITG = 32 # TODO: we don't have binarized gradient implementation yet.
# get quantized functions
f_w, f_a, f_g = get_dorefa(BITW, BITA, BITG)
def to4d(img):
return img.reshape(img.shape[0], 1, 28, 28).astype(np.float32) / 255
def prepair_data(train_img, val_img, train_lbl, val_lbl, batch_size):
train_iter = mx.io.NDArrayIter(to4d(train_img),
train_lbl,
batch_size,
shuffle=True)
val_iter = mx.io.NDArrayIter(to4d(val_img), val_lbl, batch_size)
return train_iter, val_iter
import logging
import numpy as np
import matplotlib.pyplot as plt
import pdb
from dorefa_ops import get_dorefa
from math_ops import *
from random import randint
logging.getLogger().setLevel(logging.DEBUG)
BITW = 1
BITA = 1
BITG = 6 # TODO: we don't have binarized gradient implementation yet.
# get quantized functions
f_w, f_a, f_g = get_dorefa(BITW, BITA, BITG)
def to4d(img):
return img.reshape(img.shape[0], 1, 28, 28).astype(np.float32)/255
def prepair_data(train_img, val_img, train_lbl, val_lbl, batch_size):
train_iter = mx.io.NDArrayIter(to4d(train_img), train_lbl, batch_size, shuffle=True)
val_iter = mx.io.NDArrayIter(to4d(val_img), val_lbl, batch_size)
return train_iter, val_iter
def get_lenet():
"""
original lenet
"""
data = mx.symbol.Variable('data')
# first conv layer