Chainerのサンプルコードを集めてみた（メモ）
Chainerのサンプルコードは色んな所で公開されていますが、
一箇所にまとまっている所が少ない気がしたので、個人用にまとめておきます。
ここに載ってないサンプルコードがありましたら、
コメント欄か@aonotasまで教えて下さい！！
誰得かは分かりませんが，gitにもメモ用に作りました。
github.com
適宜更新していく予定です。
Chainer OfficialのTwitterアカウントが主な情報源です。
Official wiki page (some examples list)
こちらの公式Wikiも有用です。
他のサイト様では
chainerのサンプルコードを集めてみた(チュートリアルも追加) - studylog/北の雲
がサンプルコードのまとめとしては良いかもしれません。
更新履歴
- (2016.08.19) 記事作成
- (2016.09.31) Fully Convolutional Networksを追加しました(thanks: @nizumical )
画像・動画系
DeepPose
DeepPose implementation by Chainer (@ChainerOfficial). Single stage AlexNet, 450 epoch. https://t.co/suydZkOnX6 pic.twitter.com/4JStfN83tX
— assisi (@masvsn) June 23, 2015
R-CNN
はてなブログに投稿しました #はてなブログ
— sinhrks (@sinhrks) July 5, 2015
Chainer で Deep Learning: model zoo で R-CNN やりたい - StatsFragmentshttp://t.co/0yVhIdHe89 pic.twitter.com/yA9xFqcuZx
Fast R-CNN
Fast R-CNN without caffe or GPU. Refactored from @mitmul 's code in chainer. Let's learn object detection. https://t.co/bmcElyTsrR
— Satoshi (@satoshi2373) June 14, 2016
Group-CNN
Taco Cohen published Group equivalent CNN to ICML and his impl. with #Chainer is available at https://t.co/koUVXZ9FgE. Great work, Taco!
— Chainer (@ChainerOfficial) July 20, 2016
Fully Convolutional Networks
Selective Search
Object localization & recognition using Caffe's CNN model, @AlpacaHQ 's Selective Search impl., and Chainer by @shi3z http://t.co/Ca5IVKVSol
— Chainer (@ChainerOfficial) August 10, 2015
Faster art style transfer
Faster art style transfer https://t.co/vdHuYH8vXY
— Chainer (@ChainerOfficial) June 21, 2016
illustration2vec
.@rezoolab released THE illustration2vec, which works either with Caffe, or Chainer.https://t.co/dHf8T5lx86
— Chainer (@ChainerOfficial) October 27, 2015
code https://t.co/2NHBqWyPuK
waifu2x
Image super resolution (waifu2x) impl using Chainer by @mikemoke_d. CPU-only code is disclosed http://t.co/I9jxFP5aXq pic.twitter.com/0O9vpWOhyd
— Chainer (@ChainerOfficial) July 22, 2015
artificial Gogh/Picasso machine
Wow, the artificial Gogh/Picasso machine has been reproduced with Chainer only in 1 week. Great work, @satoshi2373! https://t.co/aPSXLApUWR
— Chainer (@ChainerOfficial) September 1, 2015
Gogh (and Picasso)
Chainer-based Gogh (and Picasso) is here! https://t.co/St2BEr0fYx @mattya1089 explains how it works (translated) https://t.co/kUoG8mBXes
— Chainer (@ChainerOfficial) September 15, 2015
VGG-net
Visualization of convolution filters of VGG-net using Chainer https://t.co/O74lJPLA7n by @mitmul.
— Chainer (@ChainerOfficial) February 1, 2016
image caption generation
.@satoshi2373 reproduced the image caption generation work at CVPR'15 https://t.co/q59ea6BUv6 Chainer-based code https://t.co/xEvYom9KKc
— Chainer (@ChainerOfficial) February 16, 2016
covnet benchmark
Chainer has been added to the popular convnet-benchmarks.https://t.co/jBdzpSCBBc Thank you, Soumith(@amiconfusediam)!
— Chainer (@ChainerOfficial) January 8, 2016
NLP系
Recurrent Language Model
Recurrent Neural Network Language Model using Chainer in 222 lines by @odashi_t. https://t.co/SGIKpPHeQu
— Chainer (@ChainerOfficial) September 1, 2015
encoder-decoder MT
My neural encoder-decoder MT written by #Chainer has been updated. Please refer the latest version on Gist :)https://t.co/mKs3q5BIhD
— Yusuke Oda (@odashi_t) September 1, 2015
RNN文章学習
Japanese entrepreneur (@shi3z) tried to let Chainer replace him as a writer using his old posts w/ 1.5 million chars http://t.co/Cd6gHOpX2f
— Chainer (@ChainerOfficial) September 1, 2015
LSTM variants
A ton of LSTM variants written in @ChainerOfficial https://t.co/2zvNdFXO8H
— christopher e moody (@chrisemoody) April 21, 2016
char-embedding
Useful article by @dpcoetsee on char-embedding. Keep on! https://t.co/x1BWp1wReg
— Chainer (@ChainerOfficial) February 9, 2016
BinaryNet
The BinaryNet paper was published last week. It's Chainer implementation was given yesterday by @hillbig. Enjoy!https://t.co/5fDW2iYW1z
— Chainer (@ChainerOfficial) February 14, 2016
Deep Networks with Stochastic Depth
"Deep Networks with Stochastic Depth" in Chainer appeared just 10 days after the publication. Great work, @kchnk0! https://t.co/eiFg2mkKMu
— Chainer (@ChainerOfficial) April 13, 2016
VAE，GAN系
VAE, GAN
Chainer implementation of generative neural network models by TJ Torres (@Teejosaur). VAE, GAN and more. Awesome! https://t.co/ZPRJdSCmd7
— Chainer (@ChainerOfficial) January 27, 2016
Generative Adversarial Nets
Chainer使って1日位で作った即席Generative Adversarial Nets．画像はMNIST使って生成してみた手書き数字の一例．パラメータ適当なのでもっといいのあるはず https://t.co/lELbLycxmk pic.twitter.com/EBDX939zQZ
— Masaki Saito (@rezoolab) July 20, 2015
fauxtorgaph to make generative fonts with GANs
New blog post on using fauxtorgaph to make generative fonts with GANs in @ChainerOfficial. https://t.co/TMsOXZWw3h pic.twitter.com/7CWiS4kBkS
— TJ Torres (@Teejosaur) February 2, 2016
強化学習系
Deep Q-Network
DQNの生い立ち ＋ Deep Q-NetworkをChainerで書いた [Python] on @Qiita http://t.co/JroCiqGJfc
— Ugo-Nama (@movingsloth) July 11, 2015
Famous Deep Q-Network impl using Chainer by @movingsloth. Fun to see the Pong video. https://t.co/aOrzaWmWc7 pic.twitter.com/M0iTaPGWEq
— Chainer (@ChainerOfficial) July 22, 2015
その他
ResDrop
Test accuracy of ResDrop exceeds that of the original ResNet before it converges.https://t.co/EH25DScVzh
— Chainer (@ChainerOfficial) April 13, 2016
PredNet (http://arxiv.org/abs/1605.08104 )
Chainer impelemtation of PredNet (https://t.co/rHEGpcfshX) by @kawa_kunimasa. Thx!https://t.co/hZdlm7UJw5
— Chainer (@ChainerOfficial) June 21, 2016
Siamese
Siamese network implementation using Chainer by @mitmul. https://t.co/ulhN1ZbxHm
— Chainer (@ChainerOfficial) October 28, 2015
support vector machine（SVM）
Chainer can also be used for implementing support vector machine, of course. Thanks @mitmul.https://t.co/pQy8uO2yOo
— Chainer (@ChainerOfficial) October 27, 2015
流体計算
Fluid simulation w/ Chainer. What, @mattya1089? I like the idea of backprop for transform https://t.co/atmqRCXMeX https://t.co/kMXCzEtdx9
— Chainer (@ChainerOfficial) September 1, 2015
Convolutional Autoencoder
chainerで書いたConvolutional Autoencoderがうまくいった！！！入力画像と一層目で再構築した画像！ http://t.co/kxHuH3SF1P http://t.co/VVsmlqJhiu
— 前川ことね (@kotone_nyt) July 2, 2015
introduced Chainer as "looks extremely promising". Really?
Frameworks and Libraries for Deep Learning http://t.co/lyLbFmU02b @creative_punch introduced Chainer as "looks extremely promising". Really?
— Chainer (@ChainerOfficial) August 10, 2015
RNN to write Russian literature by modifying Chainer-user's code
.@genekogan told RNN to write Russian literature by modifying Chainer-user's code https://t.co/ZpvxG1ve27 Looks good https://t.co/4uuJm0xAc9
— Chainer (@ChainerOfficial) August 11, 2015
(speed)Caffe reinforces its dominance, Torch7 and Keras move forward, Chainer appears.
3 month later: Caffe reinforces its dominance, Torch7 and Keras move forward, Chainer appears. ping @coffeephoenix pic.twitter.com/8E3ZrJdS8j
— François Chollet (@fchollet) August 24, 2015
