Christos Tsirigotis

Including nccl multi-gpu library to PyCUDA

1

Hello there, I want to wrap NVIDIA's nccl library for multi-gpu communication collectives into Python, so as to be compatible with PyCUDA framework. I thought that it would be appropriate...

Reproducibility of MoCoV3, cifar10 with ResNet18

1

Hi there! Thank you for your work! I have tried to reproduce the reported results (~93% top-1) on CIFAR10 with ResNet18 using MoCoV3, however running the provided script with default...

Fix multi-node

3

It seems that I can execute the tests on Helios, so I will be able to fix and test the multi-node code. Nevertheless, the requirement that a multi-node/GPU job must...

Generate site for documenting Platoon and coding standards

2

I think there should be an effort to rewrite some documentation (e.g. instructions) for Platoon and generate files to have an official reference site, such the ones there are for...

Use `.item()` to collect results and losses-as-results only at an epoch's end

3

Using `.item()` to store results in `routine` call forces GPU to synchronize in order to have access at a lazy-evaluated Python number. This is suboptimal as kernel scheduling (CPU load)...

LossHandler does not handle two different sets on the same network key

4

I have come to realize by the implementation in `built_ins/gan.py` that two separate calls are done in order to update the discriminator. The first optimizer update utilizes the gan loss...

Include `seed` to argument parser

3

Hello there! I would like to make a PR which would include the following things: - [x] Add an cmd line argument for a seed. - [x] Create a setter...

Reproducing experiments of fig. 3

3

Hello there again, I am trying to reproduce the experiments of figure 3 in my own codebase, but I have not been successful in reproducing the results of the case...