Visdom: Is Visdom development dead?

Created on 30 Jul 2020  路  6Comments  路  Source: fossasia/visdom

I see a few PRs being merged from ~2 mo ago, but no active development, eg: https://github.com/facebookresearch/visdom/pull/675 has been in progress since last year.

I get it that we all have time constraints and things to focus on (I'm a prime example of this with my ~4yr old cross-GPU ML lib 馃槀). Just want to know if I should start refactoring my code to move to an alternative like wandb or not.

All 6 comments

Hi @jramapuram - I've mostly been caught up in other projects and following survey feedback on visdom there wasn't enough immediate interest for me to motivate others to help out on the work. I hope to return someday as I think it's still ideal for visualizing experiments under custom workflows, but until the requirements of my projects bring me back to using visdom I likely won't have much bandwidth to work on improvements.

Hi @JackUrb! Actually, Visdom is invaluable to my work as ML engineer and researcher, and this is probably true for many others as well. We would greatly appreciate it if you could invest more time into this project (at least, do more PR reviews). This could change ML experience for better for many developers all over the world, because it is much more functional than TensorBoard which everyone is using. ;) Would be happy to participate as well!

You could also consider submitting Visdom to GSOC, this can help to push the project forward and make it more recognisable in the community as well

@JackUrb : thanks for the update. Just to clarify again, Visdom and your work on this project is awesome and greatly appreciated. I totally understand being strewn between projects. Best of luck.

@JackUrb Just to let you know that Visdom is a central piece of my Deep Learning framework ( https://github.com/Vermeille/Torchelie/ ) because the ability to inject just any html into a text widget is an invaluable tool for structured reports, tables etc.

Visdom is an awesome product and I wanted to thank the whole team for it. Just let me know if delayed further fixing and development means I should switch to tensorboard.

Closing, as is answered in #753.

Was this page helpful?
0 / 5 - 0 ratings

Related issues

shubhamjain0594 picture shubhamjain0594  路  3Comments

amartya18x picture amartya18x  路  6Comments

dragen1860 picture dragen1860  路  6Comments

jkarimi91 picture jkarimi91  路  8Comments

wj926 picture wj926  路  7Comments