Deep learning visualizer

Neural Network Forward Pass Visualizer

See how inputs, weights, bias, activation, and prediction flow through a neural network.

TypeNeural NetworkFlowForward PassActivationReLUOutputPrediction
Input Layer(3)Hidden Layer 1(4)Hidden Layer 2(4)Output Layer(1)-0.23-0.49-0.510.480.21-0.160.230.490.51-0.48-0.210.170.490.32-0.01-0.340.080.390.500.37-0.49-0.310.020.34-0.09-0.39-0.50-0.36-0.070.360.610.570.75x1-0.30x20.50x3h1_1--h1_2--h1_3--h1_4--h2_1--h2_2--h2_3--h2_4--y_hat----
Weak weightStrong weightData flow

Forward Pass Steps

  1. Multiply inputs by weights
  2. Add bias to get weighted sum z
  3. Apply activation function
  4. Pass activations to the next layer
  5. Compute final prediction
Forward pass only computes prediction.

Training happens later with loss calculation and backpropagation to update weights.

InputForward PassLossBackprop
Forward pass = predict. Backprop = learn.