Issue with logging multiple confusion matricies

wandb, version 0.8.35 Python 3.7.7 Linux

Description

I’m trying to log a confusion matrix after each epoch using wandb.sklearn.plot_confusion_matrix(all_predictions, all_labels, class_labels, title=f'Confusion Matrix: Epoch {epoch}', normalize=True).

My expected behaviour would be multiple confusion matrices getting logged each with the appropriate title, but the actual result (shown below), is that the original matrix gets overwritten and the individual values end up written on top of each other.

[tomginsberg]

Here’s a good termporary fix

from sklearn.metrics import confusion_matrix
for epoch in range(num_epochs):
        ...
        cm = confusion_matrix(all_labels, all_predictions, normalize='pred')
        wandb.log({f'Confusion Matrix. Epoch {epoch}': wandb.plots.HeatMap(class_labels, 
                            class_labels, matrix_values=cm, show_text=True)})

Unfortunately plot labels don’t appear (maybe there’s a way to do this), but the most recent plot will be first on the media panel.

[ariG23498]

Hey @tomginsberg In the past year we’ve majorly reworked the CLI and UI for Weights & Biases. We’re closing issues older than 6 months. Please comment to reopen.