Weight Initialisation Quiz

Explore initialisation methods, solve calculations, debug code, compare gradient behaviour, and build reliable neural networks through practical adaptive challenges and visual performance reports today.

Build your quiz

Choose a mode, filter the 132-question bank, and configure scoring.

Quiz mode

Question count

Question search

Difficulty

Topic

Initializer

Activation

Framework

Layer type

Seconds per question

Passing score

Randomise questions

Randomise answer choices

Enable 25% negative marking

Use per-question timer

Adaptive difficulty

Keyboard navigation

The selected filters currently contain 132 questions.

Formula reference

Method	Variance or limit	Typical pairing
Glorot normal	Variance = 2 / (fan-in + fan-out)	Tanh, sigmoid, linear
Glorot uniform	Limit = √(6 / (fan-in + fan-out))	Tanh, sigmoid, linear
He normal	Standard deviation = √(2 / fan-in)	ReLU
He uniform	Limit = √(6 / fan-in)	ReLU
LeCun normal	Standard deviation = √(1 / fan-in)	SELU
LeCun uniform	Limit = √(3 / fan-in)	SELU
Leaky-ReLU gain	Gain = √(2 / (1 + a²))	Leaky ReLU
Dense fan values	fan-in = inputs; fan-out = outputs	Dense layers
Convolution fan values	Channels × kernel height × kernel width	Convolutional layers

Related Calculators

Neural Network Parameter Calculator Activation Function Calculator ReLU Function Calculator Tanh Function Calculator Neural Network Weighted Sum Calculator Forward Propagation Calculator Backpropagation Gradient Calculator Learning Rate Calculator Epoch and Iteration Calculator Dropout Calculator

Important Note: All the Calculators listed in this site are for educational purpose only and we do not guarentee the accuracy of results. Please do consult with other sources as well.

Weight Initialisation Quiz

Build your quiz

Formula reference

Attempt history

Quiz report

Performance charts

Strengths and revision priorities

Detailed answer review

Related Calculators