INITIALIZING SYSTEM
CNN
v1.2.1
IDLE
Data
Train Mode
Data Input
×
Upload Image
Click to upload or drag image here
Auto-resized to input dims
Presets
Current Input
Downscaler
8x8
pixels
Grayscale
Layer Inspector
Ready.
Kernel Weights
Identity
Red lines: Negative weights
Blue lines: Positive weights
Play
Step
Reset
Time Scale
1.0x
Target
Digit 0
L. Rate
0.1
Loss
0.000
Settings
×
Hyperparameters
Aesthetics
Input Volume
Dimensions (Step: 2)
Max 128x128
High Resolution: Text labels disabled
Color Mode
Grayscale (1 Layer)
RGB (3 Layers)
Convolution Layer
Kernel Filter
Identity
Sobel Horizontal
Sobel Vertical
Sharpen
Box Blur
Edge Detection
Gaussian
Padding
Valid (No Padding)
Same (Zero Padding)
Kernel Size
3x3
5x5
7x7
Stride
1
Activation
ReLU
Sigmoid
Tanh
Pooling Layer
Type
Max Pooling
Average Pooling
Fully Connected
Hidden Neurons
12
Network Depth (Layers)
1
Output Classes
10
Connection Style
Iterative (Single Step)
Cumulative (Web Mode)
View Mode
Numbers
*Forced to Voxels at >32px
Color Theme
Default (Layer Colors)
Magma (Heatmap)
Viridis
Plasma
Matrix Green
Front Image Opacity
Layer Spacing
Animation Speed