zhreshold’s gists

zhreshold / imagenet1000_clsid_to_human.txt

Created September 26, 2017 22:15 — forked from yrevar/imagenet1000_clsidx_to_labels.txt

text: imagenet 1000 class id to human readable labels

	{0: 'tench, Tinca tinca',
	1: 'goldfish, Carassius auratus',
	2: 'great white shark, white shark, man-eater, man-eating shark, Carcharodon carcharias',
	3: 'tiger shark, Galeocerdo cuvieri',
	4: 'hammerhead, hammerhead shark',
	5: 'electric ray, crampfish, numbfish, torpedo',
	6: 'stingray',
	7: 'cock',
	8: 'hen',
	9: 'ostrich, Struthio camelus',

zhreshold / README.md

Last active January 9, 2018 23:50

Super resolution example onnx

This is a demo onnx model for super resolution.

zhreshold / concat.py

Created October 4, 2017 18:57

HybridBlock of concat layer

	from mxnet.gluon import HybridBlock

	class Concat(HybridBlock):
	"""Concat operation for multiple inputs."""
	def __init__(self, dim=1, **kwargs):
	super(Concat, self).__init__(**kwargs)
	self._kwargs = {'dim': dim}

	def hybrid_forward(self, F, *args):
	return F.concat(args, name='fwd', *self._kwargs)

zhreshold / fc_bench.py

Last active October 5, 2017 18:48

FC perf benchmark

	import mxnet as mx
	import numpy as np
	from timeit import default_timer as timer

	def get_bench_net(num_hidden=10000):
	data = mx.sym.var('data')
	fc = mx.sym.FullyConnected(data, num_hidden=num_hidden)
	return fc

	num_out = 10000

zhreshold / resnet101_v2.sh

Last active October 12, 2017 23:31

MXNet Imagenet training configurations

python train_imagenet.py --data-train ~/data/train.rec --data-val ~/data/val.rec --gpus 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 --data-nthreads 32 --network resnet101_v2 --batch-size 256 --top-k 5 --model-prefix model/resnet101_v2 --min-random-scale 0.533 --max-random-shear-ratio 0 --max-random-rotate-angle 0 --max-random-h 0 --max-random-l 0 --max-random-s 0 --lr-step-epochs 30,60,90 --num-epochs 120 --rgb-std '58.395,57.12,57.375'

zhreshold / bullet.js

Last active October 14, 2017 03:44

d3.js demo

	(function() {

	// Chart design based on the recommendations of Stephen Few. Implementation
	// based on the work of Clint Ivy, Jamie Love, and Jason Davies.
	// http://projects.instantcognition.com/protovis/bulletchart/
	d3.bullet = function() {
	var orient = "left", // TODO top & bottom
	reverse = false,
	duration = 0,
	ranges = bulletRanges,

zhreshold / demo_bench.json

Last active October 14, 2017 22:23

Sortable table d3.js

	{
	"nvidia-titan-x": {
	"devices": [
	{
	"cores": "3072",
	"memory": "12GB",
	"memory_bandwith": "336.5GB/s",
	"name": "Nvidia Titan X",
	"quantity": 1
	}

zhreshold / shufflenet_1x_g3_deploy.prototxt

Created October 16, 2017 05:55

Shufflenet caffe

	name: "shufflenet"
	# transform_param {
	# scale: 0.017
	# mirror: false
	# crop_size: 224
	# mean_value: [103.94,116.78,123.68]
	# }
	input: "data"
	input_shape {
	dim: 1

zhreshold / params.sh

Last active April 27, 2018 13:00

MXNet shufflenet from https://github.com/ZiyueHuang/MXShuffleNet

python mxnet/example/image-classification/train_imagenet.py --network shufflenet --data-train ~/efs/users/joshuazz/data/imagenet/record/train_480_q95.rec --data-val ~/efs/users/joshuazz/data/imagenet/record/val_256_q90.rec --batch-size 512 --gpus 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15 --num-epochs 150 --lr-step-epochs 30,60,90 --min-random-scale 0.533 --lr 0.01 --disp-batches 100 --top-k 5 --data-nthreads 32 --random-mirror 1 --max-random-shear-ratio 0 --max-random-rotate-angle 0 --max-random-h 0 --max-random-l 0 --max-random-s 0 --model-prefix model/shufflenet  | tee ~/efs/users/joshuazz/temp/train_imagenet_logs/shufflenet.log

zhreshold / cuda9_2.sh

Last active October 26, 2018 08:24

CUDA 9 and CUDNN 7

	sudo apt update
	sudo apt-get -y install g++ freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libglu1-mesa libglu1-mesa-dev libgfortran3
	wget 'https://developer.nvidia.com/compute/cuda/9.2/Prod2/local_installers/cuda_9.2.148_396.37_linux' -O cuda92.run
	wget 'https://developer.nvidia.com/compute/cuda/9.2/Prod2/patches/1/cuda_9.2.148.1_linux' -O patch1.run
	sudo sh cuda92.run
	sudo sh patch1.run

	echo "# CUDA9.2" >> ~/.bashrc
	echo "export PATH=/usr/local/cuda/bin:\$PATH" >> ~/.bashrc
	echo "export LD_LIBRARY_PATH=/usr/local/cuda/lib64:\$LD_LIBRARY_PATH" >> ~/.bashrc

Joshua Z. Zhang zhreshold