Skip to content
Calcrivo

Parameter Count Calculator

Calculate the number of trainable parameters in a dense, convolutional, or LSTM layer.

Inputs

Input features (dense) or input channels (Conv2D/LSTM input dim).

Output units (dense), output channels (Conv2D), or hidden size (LSTM).

Only used for Conv2D layers (assumes a square kernel).

Total Parameters

131,328

Weight Parameters

131,072

Bias Parameters

256

Step by step

  1. Dense: (input × output) + output (bias)

    (512 × 256) + 256

    = 131,328

  2. Weight parameters vs. bias parameters

    131,072 (weights) + 256 (biases)

    = 131,328

How it works

Parameter count varies by layer type. A dense (fully connected) layer has params = input × output + output, where the extra output term is the bias vector. A Conv2D layer has params = kernel_h × kernel_w × in_channels × out_channels + out_channels, since the same kernel is shared across all spatial positions. An LSTM layer has four gates (input, forget, output, cell), each acting like a dense layer over the concatenated input and hidden state, giving params = 4 × [hidden × (input + hidden)] + 4 × hidden.

Formulas

Dense layer

params = input_size * output_size + output_size

input_size
Number of input features
output_size
Number of output units (neurons)

Conv2D layer

params = kernel_h * kernel_w * in_channels * out_channels + out_channels

kernel_h
Kernel height
kernel_w
Kernel width
in_channels
Input channels
out_channels
Output channels (filters)

LSTM layer

params = 4 * (hidden * (input + hidden)) + 4 * hidden

input
Input dimension
hidden
Hidden state dimension

Frequently Asked Questions

Why does Conv2D have far fewer parameters than Dense for the same input/output size?

Convolutional layers share the same small kernel across every spatial location, so the parameter count depends only on kernel size and channel counts, not on the spatial dimensions of the input.

Why does LSTM have 4x more parameters than a similarly-sized dense layer?

An LSTM cell contains four internal gates (input, forget, output, and cell/candidate), each with its own weight matrix and bias, so its parameter count is roughly four times that of one dense transformation of the same size.

Does this include normalization layer parameters?

No, this calculates only the specified layer type. BatchNorm/LayerNorm add a small number of additional parameters (typically 2 × channels) not included here.

How do I estimate a full model's parameter count?

Sum the parameter counts of every layer in the architecture. For transformer models specifically, see the Transformer Parameters Calculator for a purpose-built formula.

You might also need