bamsumit
bamsumit
@Michaeljurado24 thanks for taking this initiative. Let's start a PR (keep it as a draft). Then I can look at the PR and provide more feedback.
This should probably be looked at together with the default `build` functionality for the hierarchical process.
https://github.com/lava-nc/lava/pull/372
There are a couple of semantic issues that exist right now. 1. First, our connection processes have an argument called `num_message_bits=0` and we use `num_message_bits > 0` to say that...
BTW @Michaeljurado24, can you email me, I think it's better we have a 1:1 meeting.
Thanks @mathisrichter I'll create them, at least for inference notebooks. Training notebooks specifically require GPUs and take hours to complete. That's why we never added notebooks to the unittests. The...
This is likely due to mismatch/incompatibility between your installed pytorch cuda version and nvcc cuda version. Make sure the cuda version of both is the same.
@nskat Thanks for the PR contribution. DECOLLE is a valuable addition to lava-dl. We will review it soon and provide feedback soon. At first glance, can you remove the files...
Hi @nskat, thanks for making the changes. It looks good. __With the current state of `DECOLLELoss` implementation, we have some licensing concerns. Will you be able to reimplement it or...
Hi @nskat can you contact me at sumit.bam.shrestha(at)intel.com ?