Use of int8 and float32 as data types in model built using MXNet
In this recipe, we will see what happens when you use int8 and float32 as data types in the model built using MXNet
Recipe Objective: What happens when you use int8 and float32 as data types in the model built using MXNet?
This recipe explains what happens when you use int8 and float32 as a data type in the model building using MXNet.
A Deep Dive into the Types of Neural Networks
Table of Contents
Step 1: Importing library
Let us first import the necessary libraries. We'll import mxnet, NumPy, gluoncv, and matplotlib.pyplot.
import numpy as np
import mxnet as mx
import gluoncv as gcv
import matplotlib.pyplot as plt
Step 2: Loading The Sample Image
We'll be downloading a sample image of a dog from the URL "https://raw.githubusercontent.com/zhreshold/mxnet-ssd/master/data/demo/dog.jpg"
image_url = 'https://raw.githubusercontent.com/zhreshold/mxnet-ssd/master/data/demo/dog.jpg'
image_path='dog.jpg'
gcv.utils.download(url=image_url,path=image_path)
img = mx.image.imread('dog.jpg')
Step 3: Processing
We'll be defining a utility function that will be used for the augmentation visualization. We have loaded the image with numpy.int8 datatype because for some functions, for example, swapexe don't compute on int type, so we have converted int18 to float32.
def plot(arr):
imshow((arr.clip(0, 255)/255).asnumpy())
from matplotlib.pyplot import imshow
img = image.astype("float32")
plot(img)
What Users are saying..
Ameeruddin Mohammed
I come from a background in Marketing and Analytics and when I developed an interest in Machine Learning algorithms, I did multiple in-class courses from reputed institutions though I got good... Read More