mfkasim1

// g++ -O2 div-avx.cc -I/home/muhammad/libs/sleef/include/ -mavx -L/home/muhammad/libs/sleef/lib/ -lsleef

#include <immintrin.h>
#include <sleef.h>
#include <chrono>
#include <iostream>
#include <iomanip>
#include <cmath>
#include <complex>

mfkasim1

These are the git commands I usually use (for self-reminder), sorted from the most-used ones:

1. git status: to see the states
2. git log --oneline: to see which commit I'm at
3. git add <filename(s)>: to stage a file(s)
4. git commit -m "<Some message>": to make a commit
5. git push: push the latest update to the remote
6. git pull: pull the latest states from the remote
7. git clone <repo_url>: clone a new repository
8. git diff : to see which lines in the file that haven't been staged

mfkasim1

"""
Gradph object can be used to debug multi-level autograd for anomaly detection.
To see an example on how to use this, see the bottom of this script.
"""

import torch

class GradphNode(object):
    def __init__(self, gfn, parent_node):
        self._gfn = gfn

mfkasim1

# plot histogram with logscale in the x-axis
import numpy as np
import matplotlib.pyplot as plt
def plot_loghist(x, bins=30, **hist_kwargs):
    hist, bins = np.histogram(x, bins=bins)
    logbins = np.logspace(np.log10(bins[0]),np.log10(bins[-1]),len(bins))
    plt.hist(x, bins=logbins, **hist_kwargs)
    plt.xscale('log')

# silencing loud third-party library

mfkasim1

Supervised learning

• Create the Dataset object and test if it can load the data correctly
• Create the transformation (in DataLoader) for the Dataset
• Create a model and test if the DataLoader can feed the data to the model
• Create the training procedure

Tips

Datasets

mfkasim1

import torch

class eig(torch.autograd.Function):
    @staticmethod
    def forward(ctx, A):
        # normalize the shape to be batched
        Ashape = A.shape
        if A.ndim == 2:
            A = A.unsqueeze(0)
        elif A.ndim > 3:

mfkasim1

def bjac(gy, y):
    """
    Get the batched Jacobian matrix for vector-valued output, gy, with respect to the input, y.
    The Jacobian output will take shape of (nbatch, *gy.shape[1:], *y.shape[1:])
    """
    # reshape gy to be batched 1D vector
    ggy = gy.view(gy.shape[0], -1)
    outshape = (gy.shape[0], *gy.shape[1:], *y.shape[1:])

    nbatch, nfout = ggy.shape

mfkasim1

Check available NVIDIA

lspci | grep -i nvidia

Check NVIDIA driver version

cat /proc/driver/nvidia/version

Display log for drivers