ENJOY-NOW

0.batch-size= N   full-batch1.batch-size=1   online_learning2.natch-size=m  mini-batch

CONTACT WEIBOREFERENCESTENSORCLASS MODULEDESIGN MODULECONTAINERMULTI GPUDATASETSAVE RESTOREOPTIMIZERPARAM INITERRORSTENSORBOARDXPROFILER''' by L'''import torch import...

CONTACT WEIBOcapsule networkgenerative adversarial network (GAN)recurrent neural network (RNN)source of deep learning and mlpaper sources

CONTACT WEIBOREFERENCEIMAGE PROCESS TF.APP TENSORBOARD SERVINGSOMETHING

1. DynamicRoutingBetweenCapsules2. Matrix capsules with EM Routing-----reference---------------------https://jhui.github.io/2017/11/14/Matrix-Capsules-with-EM-routing-Capsule

---------------------reference------------------------------1. http://blog.aylien.com/introduction-generative-adversarial-networks-code-tensorflow/

ICCV, CVPR : http://wwww.cv-foundation.org/NIPS: https://papers.nips.cc/ICML JMLR AISTAS http://jmlr.org/processings/papers/CFF  http://www.cff.org.cn/http://axive.org

--------------------reference-----------------------0. https://github.com/rbgirshick/py-faster-rcnn1. http://blog.youkuaiyun.com/samylee/article/details/512017442. http://blog.youkuaiyun.com/hgd1010101/arti

about rnn

----------attention-----------http://blog.youkuaiyun.com/shenxiaolu1984/article/details/51582185

------------------0------------------------------#/usr/bin/bahIMAGE_ROOT=xxxrm -rf train.txtfind IMAGE_ROOT/image/ -name *xx.jpg | cut -d '/' -f3 | sed "s/$/ 1/" >> train.txtfine IMAGE_ROOT/

1.  Hashing图像检索源码及数据库总结  http://blog.youkuaiyun.com/willard_yuan/article/details/29806755

1. http://dustintran.com/blog/

import numpy as npimport randomdef sigmoid(z):    return 1.0/(1.0 + np.exp(-z))def sigmoid_prime(z):    return sigmoid(z)*(1.0-sigmoid(z))class Net(object):    def __init__(self,

 1. 单机多卡数据并行```model = xxx  losses = torch.nn.parallel.data_parallel(model, inputs=(), device_ids=[], dim=x)  # functional style```2. pytorch-1.0 distributed2.1 单机多卡 2.2 多级多卡 ----...