zhreshold

Class	Pytorch	MXNet Gluon
Dataset holding arrays	torch.utils.data.TensorDataset(data_tensor, label_tensor)	gluon.data.ArrayDataset(data_array, label_array)
Data loader	torch.utils.data.DataLoader(dataset, batch_size=1, shuffle=False, sampler=None, batch_sampler=None, num_workers=0, collate_fn=<function default_collate>, drop_last=False)	gluon.data.DataLoader(dataset, batch_size=None, shuffle=False, sampler=None, last_batch='keep'(discard, rollover), batch_sampler=None, batchify_fn=None, num_workers=0)
Sequentially applied sampler	torch.utils.data.sampler.SequentialSampler(data_source)	gluon.data.SequentialSampler(length)
Random order sampler	torch.utils.data.sampler.RandomSampler(data_source)	gluon.data.RandomSampler(length)

zhreshold

import argparse
import mxnet as mx

parser = argparse.ArgumentParser('test')
parser.add_argument('-j', '--num-workers', default=4, type=int, dest='num_workers')
args = parser.parse_args()

dataset = mx.gluon.data.vision.MNIST()

loader = mx.gluon.data.DataLoader(dataset, 32, True, num_workers=args.num_workers)

zhreshold

git clone https://github.com/zhreshold/mxnet -b model_zoo
cd mxnet/example/gluon
sudo -H pip install -U mxnet-cu90
python image_classification.py --dataseet --train-data ~/efs/users/joshuazz/data/imagenet/record/train_480_q95.rec --val-data ~/efs/users/joshuazz/data/imagenet/record/val_480_q95.rec --batch-size 64 --num-gpus 4 --epochs 120 --lr 0.1 --mode hybrid --model resnet50_v2 --log-interval 200

zhreshold

try

sudo nvidia-smi --gpu-reset -i 0

or

sudo fuser -v /dev/nvidia*

sudo fuser -k /dev/nvidia*

zhreshold

import mxnet as mx
from mxnet import gluon

dataset = gluon.data.vision.MNIST()
loader = gluon.data.DataLoader(dataset, 34, last_batch='rollover', num_workers=8)

ctx = [mx.gpu(i) for i in range(2)]
for e in range(10):
    for i, batch in enumerate(loader):
        data = gluon.utils.split_and_load(batch[0], ctx_list=ctx)

zhreshold

import os
import argparse
import shutil
import time
import logging
import numpy as np
import mxnet as mx
from mxnet import gluon
from mxnet import autograd
from mxnet.gluon import nn

zhreshold

import argparse
import logging
import random
import time
import mxnet as mx
from mxnet import nd
from mxnet import image
from mxnet import gluon
from mxnet import autograd
import numpy as np

zhreshold

import mxnet as mx
import numpy as np
import cv2

def rotate_ndarray(arr, degree):
  theta = np.pi / 180 * degree
  cos = np.cos(theta)
  sin = np.sin(theta)
  loc = mx.nd.array([[cos, sin, -sin, cos, 0, 0]])
  s = 200

zhreshold

import mxnet as mx


class WeightedLogisticRegressionOutput(mx.operator.CustomOp):
    """

    """
    def __init__(self, beta=0.5, lower=0.3, upper=0.7):
        self._lower = lower
        self._upper = upper

zhreshold

mkdir -p n01440764	
mkdir -p n01443537	
mkdir -p n01484850	
mkdir -p n01491361	
mkdir -p n01494475	
mkdir -p n01496331	
mkdir -p n01498041	
mkdir -p n01514668	
mkdir -p n01514859	
mkdir -p n01518878