Marcos Reinan de Assis Conceição marcosrdac

Multi-node-training on slurm with PyTorch

What's this?

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.

Mutt + isync multiple accounts Unfinished

This is the setup that I use for mutt, I have two google domain account (read as gmail) and an institution where I work and study account. This means I have two gmail accounts and one outlook 365 account that i want to sync and read via mutt.

I want to store all my email locally as I travel a lot and will be in countries without easy internet access. For this I use mbsync (iSync). As it can handle multiple account types easily and efficently.

The setup works this way

[Remote Mail Servers] <= mbsync => [Local Mail Folders]

Semantic Commit Messages

See how a minor change to your commit message style can make you a better programmer.

Format: <type>(<scope>): <subject>

<scope> is optional

Example

With this wiki2html.sh bash script and pandoc program, you can convert markdown to html.

Usage: In the vim list section of the .vimrcfile, include options:

let g:vimwiki_list = [{'path': ‘your_wiki_place',
  \ 'path_html': ‘wiki_html_location’,
  \ 'syntax': 'markdown',
 \ 'ext': '.md',

To remove a submodule you need to:

Delete the relevant section from the .gitmodules file.
Stage the .gitmodules changes git add .gitmodules
Delete the relevant section from .git/config.
Run git rm --cached path_to_submodule (no trailing slash).
Run rm -rf .git/modules/path_to_submodule (no trailing slash).
Commit git commit -m "Removed submodule "
Delete the now untracked submodule files rm -rf path_to_submodule

	{
	description = "veloxchem-hpc";
	inputs = {
	nixpkgs.url = "github:NixOS/nixpkgs/nixpkgs-unstable";
	flake-utils.url = "github:numtide/flake-utils";
	pypi-deps-db = {
	url = "github:DavHau/pypi-deps-db/a8ea7f774b76d3b61237c0bc20c97629a2248462";
	flake = false;
	};
	mach-nix = {

	{ config, pkgs, lib, ... }: {
	imports = [
	(import <nixpkgs/nixos/modules/system/etc/etc.nix>)
	];

	options.system.build.etc = lib.mkOption { type = lib.types.package; };
	options.system.activationScripts.etc =
	lib.mkOption { type = lib.types.unspecified; };

	config = {

	" Specify a directory for plugins
	call plug#begin('~/.vim/plugged')

	Plug 'neoclide/coc.nvim', {'branch': 'release'}
	Plug 'scrooloose/nerdtree'
	"Plug 'tsony-tsonev/nerdtree-git-plugin'
	Plug 'Xuyuanp/nerdtree-git-plugin'
	Plug 'tiagofumo/vim-nerdtree-syntax-highlight'
	Plug 'ryanoasis/vim-devicons'
	Plug 'airblade/vim-gitgutter'

	# general commands
	alias t="task"
	alias ta="task add"
	alias td="task done"
	alias t-="task delete"
	alias tm="task modify"
	alias tu="task undo"
	alias tv="vit rc.alias.next=list"
	alias tw1="task modify wait:1d"
	# alias tw2="task modify wait:2d"

	import numpy as np
	from numpy.fft import fft2, ifft2


	def overlapadd2(Amat, Hmat, L=None, Nfft=None, y=None, verbose=False):
	"""
	Fast two-dimensional linear convolution via the overlap-add method.

	The overlap-add method is well-suited to convolving a very large array,
	`Amat`, with a much smaller filter array, `Hmat` by breaking the large