- A simple note for how to start multi-node-training on slurm scheduler with PyTorch.
- Useful especially when scheduler is too busy that you cannot get multiple GPUs allocated, or you need more than 4 GPUs for a single job.
- Requirement: Have to use PyTorch DistributedDataParallel(DDP) for this purpose.
- Warning: might need to re-factor your own code.
- Warning: might be secretly condemned by your colleagues because using too many GPUs.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
<!-- | |
Inspired by | |
https://gist.github.com/mndza/74736feb073fc6b65334#file-gistfile1-html | |
https://github.com/denysvitali/megadecrypter | |
https://gist.github.com/artjomb/7ef1ee574a411ba0dd1933c1ef4690d1 | |
About Mega Links | |
http://megadownloaderapp.blogspot.com/2013/03/explaining-mega-links.html | |
--> | |
<!DOCTYPE html> | |
<html> |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#Given a video of any aspect ratio this script will extract center cropped thumbnails at 299x299. | |
#Useful for gathering image training data. | |
#Assumes ffmpeg 3.3.2 | |
#If you get an error about exact that means you are using an older version of ffmpeg | |
#Simply remove the :exact=1. | |
#This will make it work, but may cause the output size to be off by a pixel | |
#Script assumes you are downsampling. | |
#!/bin/bash |
Use Python to:
- send a plain text email
- send an email with attachment
- receive and filter emails according to some criteria
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import sys | |
sys.path.insert(0, 'python/') | |
import caffe | |
from caffe.proto import caffe_pb2 | |
net_param = caffe_pb2.NetParameter() | |
net_str = open('lenet_iter_5000.caffemodel', 'r').read() | |
net_param.ParseFromString(net_str) | |
print net_param.layer[0].name # first layer |