Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The LossGraphNode creates a visual graph of training loss values over time and saves it as an image file. It takes loss data from training processes and generates a line chart showing how the loss changes across training steps. The resulting graph includes axis labels, min/max loss values, and is automatically saved to the temporary output directory with a timestamp.

Inputs

ParameterData TypeRequiredRangeDescription
lossLOSSYesMultiple options availableThe loss data containing loss values to plot (default: empty dictionary)
filename_prefixSTRINGYes-The prefix for the output image filename (default: “loss_graph”)
Note: The loss parameter requires a valid loss dictionary containing a “loss” key with loss values. The node automatically scales the loss values to fit the graph dimensions and generates a line plot showing the loss progression over training steps.

Outputs

Output NameData TypeDescription
ui.imagesIMAGEThe generated loss graph image saved to the temporary directory