Coding Sandbox

Run Python directly in your browser. No installation required. Experiment with DCGAN parameter counting, VAE loss, KLD, diffusion models, and more.

Templates:

# Advanced AIML Sandbox - Run Python in your browser!
# Try these examples:

# 1. DCGAN Parameter Counting
def conv_params(k, c_in, c_out, include_bn=True):
    base = k*k*c_in*c_out + c_out
    return base + (2*c_out if include_bn else 0)

k = 3
conv1 = conv_params(k, 1, 25, include_bn=False)  # First layer, no BN
conv2 = conv_params(k, 25, 50, include_bn=True)
fc = 50*2 + 2
print(f"DCGAN Discriminator params: {conv1 + conv2 + fc}")

# 2. VAE Loss
import math
def vae_loss(x, x_recon, mu, sigma_sq):
    recon = (x - x_recon)**2
    kl = 0 if (mu == 0 and sigma_sq == 1) else 0.5 * (mu**2 + sigma_sq - math.log(sigma_sq) - 1)
    return recon + kl
print(f"VAE loss (x=5, x̂=4, μ=0, σ²=1): {vae_loss(5, 4, 0, 1)}")

Output

Loading Python runtime...

Tips

Use the template buttons to load example code for each topic.
NumPy and other packages can be loaded via micropip if needed.
First run may take a few seconds while Python initializes.