blazespinnaker
blazespinnaker
For sure,- visualization is an excellent use case of why the **hooks** are very important. Maybe changing the title or opening a new issue might be a good idea so...
@contramundum53 your points about backwards compatibility are well considered. Honest question - can you enumerate some scenarios where it's possible to guarantee forward compatibility visualization but not a stable backing...
@nzw0301 Who can we reach out to get consensus on this? Even if the answer is that we just have to fork the library, at least it's an answer. This...
@contramundum53 Thanks, not sure I agree with the conclusions but the answer is actually very helpful. A forking we will go :) My two bits tho, fwiw: - There are...
ah, got it, yeah optuna studies shows a different help, which is why i missed it. Do you think there should be an option, like --show-user-attrs? or what would be...
My concern might be that if we do it by default it will make the table view mostly unreadable. Currently it's quite clean, at least for single objective trials. An...
That's fine. I was also hoping for some more discussion.
@nzw0301 Sure, though I'm a bit unsure by the above. Does this mean the PR should include a show user attr flag? Without it, I believe workflow will be broken...
Some further context: I did use pkill a couple of times on the processes running and restarted the study optimizations
Further note - Restarting optuna-dashboard seems to have fixed it.