The optimizer function runs when another guess is made. It tries to ==minimise the loss based on the loss function==.

Example in TensorflowJS

The adam optimiser is effective and requires no configuration.

// Prepare the model for training.
  optimizer: tf.train.adam(),
  loss: tf.losses.meanSquaredError,
  metrics: ['mse'],

Example in Python

Example using mean_squared_error for the loss and stochastic gradient descent (sgd) for the optimizer.

# add loss and optimizer to model
model.compile(optimizer='sgd', loss='mean_squared_error')