utils.visualization

Visualization module for creating training plots from the training statistics.

soulsai.utils.visualization.save_plots(x: np.ndarray, ys: list[np.ndarray], xlabel: str, ylabels: list[str], path: Path, N_av: int = 50)

Plot and save the training progress dashboard.

Stats are smoothed by computing the mean over a running window. Confidence intervals are given by computing the standard deviation over a running window.

Parameters:

• x – Data for the x-axis.

• ys – Data for the y-axis. For each y, a running mean and standard deviation is computed.

• xlabel – Label for the x-axis.

• ylabels – Labels for the y-axis.

• path – Save location for the figure.

• N_av – Moving average window for smoothing the plot.