Updating run summary parameters after finish/sync


I’d like to add a key/val pair (in the run summary) to a finished run, so I can do some grouping in plots. This is in recent runs, but some older experiments didn’t log it. Is it sufficient to do the following without disrupting any information that’s already uploaded (and do I need to run run.finish()?):

wandb.init(id=run_id, resume=True)


Hi Josh,

Yes, you can add a key/val pair to a finished run by using the W&B Public API. You can use the following code to update the summary attribute of an existing W&B Run after training has completed:

import wandbapi = wandb.Api()run = api.run("username/project/run_id")run.summary["<pair_name>"] = <pair_val>run.summary.update()

You do not need to call wandb.finish() as the run is already finished.


Thanks, that works a treat!

No problem! Always happy to help :slightly_smiling_face:

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