Sorry for being bit delay with the lesson timeline.
I created a MNIST basemodel for 3 and 8 and got ~89% accuracy.
I used the main MNIST dataset.
There’s nothing unexpected about that. But when I looked at images, they are not black and white. See:
Could you someone explain me why is that?
Thanks.
I think it is normal when you plot a 2D image, I always see it with plt.imshow 2d image. You can, firstly, check the dimension of your data (Which I think is 2). Secondly, convert it to 3d img with np.stack((your-img)x3times), or you can check if there is a gray colormap for it. But this is just for visualizing the image, you can ignore it if what you concentrate is the model,
Hope that helps
add this line:
matplotlib.rc('image', cmap='Greys')
Oh yes. Thanks a lot. That was the issue.
I missed that.
Thanks for the explanation.