filmov
tv
MNIST Autoencoder 03 - TF Live Coding Tiled Blend Between Numbers Python Jupyter Notebook TensorFlow

Показать описание
Live Coding in Jupyter Notebook using TensorFlow Keras in Python. An autoencoder is trained, latent space is calculated, and from the latent spaces I blend between different numbers.
0:00 - Description of what I'm going to do
0:28 - Showing similar code from previous project
1:37 - Copying chunks, describing difference
7:00 - Transitioning to the direct coding
10:45 - More detail of what I'm doing now vs next
12:12 - First try, not working yet
13:35 - Discover negative numbers in my output
19:26 - I think I found the cause of negative numbers
20:00 - Retraining
20:55 - Fixed negatives, image still blank
22:00 - Initial image draws now
26:40 - Initial video exported
27:32 - Starting to code the full grid
27:43 - Starting grid of latent spaces
35:57 - Looks like it worked. NOPE
37:04 - I think I found the problem
39:40 - Results
41:45 - Code Explanation
0:00 - Description of what I'm going to do
0:28 - Showing similar code from previous project
1:37 - Copying chunks, describing difference
7:00 - Transitioning to the direct coding
10:45 - More detail of what I'm doing now vs next
12:12 - First try, not working yet
13:35 - Discover negative numbers in my output
19:26 - I think I found the cause of negative numbers
20:00 - Retraining
20:55 - Fixed negatives, image still blank
22:00 - Initial image draws now
26:40 - Initial video exported
27:32 - Starting to code the full grid
27:43 - Starting grid of latent spaces
35:57 - Looks like it worked. NOPE
37:04 - I think I found the problem
39:40 - Results
41:45 - Code Explanation