search

sksq96 / pytorch-summary

Model summary in PyTorch similar to `model.summary()` in Keras
MIT License
3.98k stars 412 forks source link

FYI TO ISSUE SUBMITTERS #124

Open TylerYep opened 4 years ago

TylerYep commented 4 years ago

Since this project hasn't been worked on in a few months, I decided to do a rewrite of the fixing a lot of the changes.

The changes are available on my fork and is installable using: pip install torchinfo

https://github.com/TylerYep/torchinfo No changes to any of your code! It all works the same way but also handles specifying columns and handles LSTMs and RNNs as input, and incorporates @sangyx branching structure as output.

I plan on merging these two projects in the coming months (waiting on @sksq96 ), however if you need solutions to that problem soon you can just use my fork in the meantime.

Naireen commented 4 years ago

HI @TylerYep, I've been one of the maintainers of this project, and now currently have some time to be able to devote to this. If you're interested, I can work with you to merge the projects.

TylerYep commented 4 years ago

That would be great! What would be the best way to discuss this?

Naireen commented 4 years ago

Slack is the main platform comes to mind, though I'm open to any alternative suggestions!

TylerYep commented 4 years ago

Sounds good - could you send me a link? My email is on my profile page.

Naireen commented 4 years ago

Sent, let me know if you receive it!

nkrot commented 3 years ago

any news?

TylerYep commented 3 years ago

I never received write access to this project and, based on the implementation, am not too enthusiastic about handling backwards-incompatible changes in the new rewrite. Simply releasing a new version without a proper plan could break existing code in over 1400 codebases (based on GitHub). Additionally, torch-summary has deviated enough that I no longer regard it as a fork of this project - there is almost no overlap in code anymore.

I would recommend using my version if you want bug fixes and slightly more maintainability for the foreseeable future. If we gain a solid team of contributors we can revisit the merge.