Very Slow UI - How to bulk delete charts?

I’ve been using W&B for 1.5 years and this is the first time I am running into a problem. This problem is very serious and I can’t access any of my runs.

The UI is very slow because there are 129 charts that were created automatically in my charts section of my project.

I am confused because I did not manually create these 129 charts, they were automatically created when I completed my run.

It takes over 5 minutes to open a run from the table, then when I go to the charts section, it takes almost 10 minutes to delete 1 of the 129 charts.

The whole entire UI is completely freezing and slow because of this. When I delete the charts section, all 129 charts go to the Hidden Charts section, and I cant delete that section at all.

Hi @japi , thank you for writing in. This is currently the intended behavior for after deleting charts. The charts when deleted from your workspace will populate in the hidden panels section of the UI. This is being looked at by our team and I will provide an update once I hear back. In the meantime, I have asked the backend team to review your workspace to see what can be done to assist you.

Hi @japi

I appreciate your patience on this while I verified with App team the functionality of deleted graphs. The behavior your are seeing is as designed. When a user “deletes” a panel, it simply gets moved to the “Hidden Panels” section. You should keep this section closed so that those panels don’t affect workspace performance. That hidden panel section functions as a sort of trash can, but you can’t delete charts from that section entirely.

In terms of your new project, how are you logging your data? Can you please provide a brief code example for us to review. We might be able to reduce the number of charts logged based on how you are setting up your experiments.

Hi Mohammad, the new project is actually running fine.

I am logging the metrics from the sci-kit learn classification_report for my testSets

classification_report1 = classification_report(allLabels,allPredixs, output_dict=True)

prediction_time = time.perf_counter()

accuracy = accuracy_score(allLabels,allPredixs)


api = wandb.Api()

run =

print(“ALL KEYS”,run.summary.keys())

for key in classification_report1[‘non_shaky’]:

run.summary[f’testSet_{testSetName[testSetIndex]}.classification_report.non_shaky.{key}'] = classification_report1[‘non_shaky’][key]

print(f’testSet_{testSetName[testSetIndex]}.classification_report.non_shaky.{key}: ',classification_report1[‘non_shaky’][key])

for key in classification_report1[‘shaky’]:

run.summary[f’testSet_{testSetName[testSetIndex]}.classification_report.shaky.{key}'] = classification_report1[‘shaky’][key]

print(f’testSet_{testSetName[testSetIndex]}.classification_report.shaky.{key}: ',classification_report1[‘shaky’][key])

run.summary[f"testSet_{testSetName[testSetIndex]}.Prediction"] = prediction_time-start_time

run.summary[f"testSet_{testSetName[testSetIndex]}.Test Error Rate"] = round((1-accuracy)*100, 2)



I created a new project under the name “japi-tuc” however I ran into the same problem, there are only 44 runs however almost 130 charts were created.

I understand that when deleting charts they are sent to the hidden_charts section.

However even when I delete the chart and they are in the hidden_charts section, it seems almost impossible to load the project table still.

It is taking over 6 minutes to load the project table with just 40 runs.

Please let me know if there is a way around this !

Kind regards,


Hi @japi , that number of charts shouldn’t impact your workspace performance, especially as most are in the hidden panels section. Try to:

  • Clear browser history and cache
  • Access the project in a different web browser

Does the issue persist?

Please watch the screen recording I sent to your email where I successfully accessed/browsed your new project without running into the behavior you are experiencing. When time permits, provide a screen recording of the behavior your are experiencing while displaying the browser console tab (this may provide us additional clues into what may be occurring). I appreciate your patience as we continue to look into this.

Hi @japi since we have not heard back from you we are going to close this request. If you would like to re-open the conversation, please let us know!

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.