-
Notifications
You must be signed in to change notification settings - Fork 5
New feature additions to core Bonsai.ML.Torch package
#77
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
New feature additions to core Bonsai.ML.Torch package
#77
Conversation
…tensors by decomposing along a dimension
…native libtorch method
…odes (enabled grad, inference, no grad)
…rn empty string rather than empty tensor
…n is only happening on subscription and not before that
…` to avoid confusion with what is considered "native" in this case (is .NET native or torch native?)
…ors true function (splitting along dimension)
c7c4771 to
26900fd
Compare
Co-authored-by: glopesdev <g.lopes@neurogears.org>
…lied it to `CreateTensor` and `Diagonal` operators
…d values of range
glopesdev
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@ncguilbeault Looks good to me, I just made comments on a few missing XML docs in case you want to fill them in, otherwise I am happy to merge this.
|
@ncguilbeault All looks good to me, I would just look into the few remaining unresolved conversations above, especially the ones around |
… TorchSharp's .NET specific serialization method
…iables are set explicitly
glopesdev
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
All looks great now, just a few missing bits of documentation, but we could also leave them for the next PR, if easier.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I noticed there are also some missing XML doc strings in other packages, and this was already flagged in #54 anyway, so I suggest we do a separate PR later to resolve this. For now I'm happy to merge this branch as-is.
Overview
This PR adds a few new operators and enhancements to the
Bonsai.ML.Torchpackage. Major additions include new operators for buffering, deconstructing, and printing tensors, as well as operators for controlling gradient and inference modes. There are also improvements to file format support and device initialization, plus a new base class for automatic tensor type conversion.