Weights
Definition: Weights are the numerical values of a neural network, learned during training, that set the strength of connections between neurons.
They are what get adjusted to lower the model's error. Releasing a model's weights means handing over the core of what it learned, hence the notion of open-weight models.