Similar to this thread, i ran alot of sweeps and have currently 120 runs. I am grouping them by 3 fields so that i receive 10 x 4 x 3 runs (grouped by the fields). Now when i try to visualize the val_f1 score line for every group (model_arch), one is missing. As described in the linked thread, there should be an option to select the ‘Sampled Runs’ in the Grouping Tab, however i do not have this feature, but why? The fact that currently 9 / 10 models are shown, where each has 12 runs, this would sum up to 108 runs. However (i guess) 100 runs is the max to show by default. Thus, i cannot be sure that all the runs of my first 9 models are contained and that the mean line plot is valid.
I need all my model runs to be visualized correctly within one plot, otherwise i cannot compare them. Am I overseeing something? My only other option would be to download all the data and plot it by myself laboriously.
I wanted to let you know that our engineering team is actively working on this issue. We’re aware of it and it’s on our radar.
I’ve linked your ticket to our internal engineering ticket to ensure your specific case gets the attention it deserves. This also means you’ll receive updates directly as we make progress.
We appreciate your patience while we work on a solution. If you have any questions or need any additional information in the meantime, please don’t hesitate to ask.
Thanks for bringing this to our attention, and for being a valued Weights & Biases user!
