czotti

# prepare docker on big hard drive
mkdir /home/docker
cd /var/lib/
ln -s /home/docker .

# Update system
pacman -Syu --noconfirm

# install tools
pacman -S --noconfirm vim emacs docker nginx iptables

czotti

#include <vector>
#include <random>
#include <set>
#include <map>
#include <algorithm>
#include <chrono>
#include <cstdlib>
#include <iostream>
#include <iomanip>

czotti

def get_updates(self, learning_rate, grads, lr_scalers=None):
        """
        Provides the symbolic (theano) description of the updates needed to
        perform this learning rule. See Notes for side-effects.
        Parameters
        ----------
        learning_rate : float
            Learning rate coefficient.
        grads : dict
            A dictionary mapping from the model's parameters to their

czotti

import nibabel as nib

def read_file(filename):
  """
  Args:
  -----
    filename(str): path of your file
    
  Output:
  -------

czotti

czotti

[package]
name = "approx"
version = "0.1.0"
authors = ["Clément Zotti <clement.zotti@imeka.ca>"]

[dependencies]
ndarray = ">=0.10.12,<0.12.0"
num = "0.1"

[dependencies.approx]

czotti

#!/usr/bin/env nextflow

modes = [0, 1, 2, 3, 4]
process gpu_output {
    echo true

    script:
    """
    echo \$CUDA_VISIBLE_DEVICES
    """

czotti

[package]
name = "approx"
version = "0.1.0"
authors = ["Clément Zotti <clement.zotti@imeka.ca>"]

[dependencies]
ndarray = ">=0.10.12,<0.12.0"
num-complex = "0.2.0"

[dependencies.approx]

czotti

************************************** TENSOR SIZE (1, 2, 256, 256, 256) **************************************
Iter 1/5, loss: 1.1791579723358154
-----------------------------------  ---------------  ---------------  ---------------  ---------------  ---------------
Name                                        CPU time        CUDA time            Calls        CPU total       CUDA total
-----------------------------------  ---------------  ---------------  ---------------  ---------------  ---------------
convolution                            1993651.502us    1992817.026us                4    7974606.009us    7971268.105us
add                                       1618.376us       6155.640us                4       6473.504us      24622.559us
torch::autograd::AccumulateGrad           6500.853us          7.487us                6      39005.119us         44.922us
is_floating_point                            9.718us          7.812us                1          9.718us          7.812us
cudnn_convolution_backward    

czotti

import torch
import torchvision

model = torchvision.models.resnet18(pretrained=True)

def bn_to_in_inplace(model):
    for name, module in model.named_children():
        if isinstance(module, torch.nn.BatchNorm2d):
            setattr(model, name, torch.nn.InstanceNorm2d(module.num_features))
        else: