Also known as: parameter
Parameters are internal model values, such as weights and biases, that are learned or adjusted during training. They determine how a model transforms inputs into outputs and are distinct from hyperparameters, which configure the learning process.
Internal variables that a model uses to determine how the model processes input data and how it generates outputs. During training, you can adjust model parameters, such as weight and bias, to optimize the model's performance. During inference, you can influence the model's output through various prompting parameters, which doesn't directly change the learned model parameters. Key prompting parameters include temperature, topP, topK, and maxOutputTokens.
The weights and biases that a model learns during training. For example, in a linear regression model, the parameters consist of the bias (b ) and all the weights (w~1~, w~2~, and so on) in the following formula: In contrast, hyperparameters are the values that you (or a hyperparameter tuning service) supply to the model. For example, learning rate is a hyperparameter.