-
## TL:DR
We currently allow for stealing of `split-shuffle` tasks. That's a really bad idea...
## Intro
Our distributed shuffle algorithm currently employs three kinds of tasks which are repe…
-
Discussion within issue #13 has moved farther away from the original intent of that issue.
This issue represents the ability to extend the Source and Target Clients by using ServiceLoader to find a…
-
I noticed that the data was not shuffled correctly while training. Seems that there should be `set_epoch` added when distributed training is used. See here https://github.com/pytorch/examples/blob/fe8…
-
Flamingo model initialized with 23461888 trainable parameters
╭───────────────────── Traceback (most recent call last) ──────────────────────╮
│ /home/jovyan/taoheng/work/Multimodal-GPT/mmgpt/tr…
-
Converting from NetCDF to zarr will enable us to make use of distributed workflows in Pangeo. Can we explore converting the MAR data? A few options include:
* using the [to_zarr](http://xarray.pyda…
-
**Describe the bug**
work queue executor does not have type annotations. This means that it always passes `make mypy` when testing the code, even when incorrect types are used.
**To Reproduce**
1…
-
We will endorse the use of Parallax with our Distributed OS( NeuroKernel ), however, the API documentation especially the documentation for each method is largely incomplete. We would really appreciat…
ghost updated
8 years ago
-
I'm running Cuckoo 2.0.6 on 3 separate Ubuntu 18.04 nodes trying to set up cuckoo distributed. I've gotten Cuckoo up and running on all of them with the uwsgi/nginx deploy. I'm using a mysql database …
-
Kerberos ticket verification fails in MapR distribution of hadoop. We need to handle this case separately in `Driver.java` such that skein can work on MapR distribution of hadoop as well. This will op…
-
Is there a working example of creating the vtu file to be read into VTK with data on multiple ranks? I was able to get the vtx reader to work with all data on a single rank, but not with distributed d…