Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
mxnet-on-aws-batch
Search
ryo nakamaru
April 01, 2017
Programming
0
780
mxnet-on-aws-batch
JAWS-UG HPC * AI @ 2017.03.031
ryo nakamaru
April 01, 2017
Tweet
Share
More Decks by ryo nakamaru
See All by ryo nakamaru
AWSで楽をするサービスメッシュ入門/appmesh-trial
pottava
1
1.4k
reinforce-2019-recap-lt
pottava
2
4.1k
ScaleShift-jp-2019-summer
pottava
1
200
Firecracker とは何か/what is Firecracker
pottava
13
5.3k
ハイブリッド並列 on Kubernetes/hybrid-parallel-program-on-kubernetes
pottava
1
410
AWS Fargate + Code 兄弟で始める継続的デリバリー / Continuous Delivery with AWS Fargate and Code brothers
pottava
12
3.1k
Singularity と NVIDIA GPU Cloud で作る ハイブリッド機械学習環境の構築 / Building a hybrid environment for Machine Learning with Singularity and NGC
pottava
3
1.3k
明日から始めるちょい足し λ / get-started-with-aws-lambda
pottava
4
2.4k
NGC と Singularity によるハイブリッド機械学習環境 / A hybrid environment for Machine Learning with NGC and Singularity
pottava
0
470
Other Decks in Programming
See All in Programming
AWSマネコンに複数のアカウントで入れるようになりました
yuhta28
2
160
チームリードになって変わったこと
isaka1022
0
190
CI改善もDatadogとともに
taumu
0
110
最近のVS Codeで気になるニュース 2025/01
74th
1
250
【PHP】破壊的バージョンアップと戦った話〜決断と説得
satoshi256kbyte
0
120
2,500万ユーザーを支えるSREチームの6年間のスクラムのカイゼン
honmarkhunt
6
5.1k
Formの複雑さに立ち向かう
bmthd
1
720
テストをしないQAエンジニアは何をしているか?
nealle
0
130
Rails アプリ地図考 Flush Cut
makicamel
1
110
[JAWS-UG横浜 #80] うわっ…今年のServerless アップデート、少なすぎ…?
maroon1st
1
170
Djangoアプリケーション 運用のリアル 〜問題発生から可視化、最適化への道〜 #pyconshizu
kashewnuts
1
230
動作確認やテストで漏れがちな観点3選
starfish719
6
1k
Featured
See All Featured
Designing Dashboards & Data Visualisations in Web Apps
destraynor
231
53k
Dealing with People You Can't Stand - Big Design 2015
cassininazir
366
25k
Large-scale JavaScript Application Architecture
addyosmani
510
110k
We Have a Design System, Now What?
morganepeng
51
7.4k
GitHub's CSS Performance
jonrohan
1030
460k
Building Adaptive Systems
keathley
40
2.4k
Unsuck your backbone
ammeep
669
57k
Done Done
chrislema
182
16k
Building Better People: How to give real-time feedback that sticks.
wjessup
366
19k
A better future with KSS
kneath
238
17k
Bash Introduction
62gerente
610
210k
Easily Structure & Communicate Ideas using Wireframe
afnizarnur
193
16k
Transcript
MXNet on AWS Batch JAWS-UG: HPC #9 & AI #5
߹ಉษڧձ @ 2017.03.31
@pottava SUPINF Inc.
• AWS Batch ͷಛ • MXNet on AWS Batch σϞ
• ߏɺϙΠϯτ • ϋϯζΦϯͷ͝հ ͓
AWS Batch ͷಛ
Պֶٕज़ܭࢉɾϋΠύϑΥʔϚϯείϯϐϡʔςΟϯά ༻్ͰਅՁΛൃش͢Δɺେنͳεέʔϧɺδϣϒͷґ ଘఆ͕ٛՄೳͳϚωʔδυฒྻࢄॲཧج൫ɻ AWS Batch
͢Ͱʹ Black Belt ͷࢿྉ͕ެ։͞Ε͍ͯ·͢ɻ AWS Batch http://aws.typepad.com/sajp/2017/02/aws-black-belt-online-seminar-aws-batch.html
ࢲϢʔβࢹͰݱঢ়Λ·ͱΊ·ͨ͠ɻ AWS Batch http://qiita.com/pottava/items/d9886b2e8835c5c0d30f
MXNet on AWS Batch
σϞ
σΟʔϓϥʔχϯάΛར༻ʢLSTM with MXNetʣ AWS നॻɺPDF 27 ϑΝΠϧΛֶशσʔλʹར༻ ग़ͩ͠ͷ୯ޠ͔ΒɺͦΕΒ͍͠ޙଓͷจষΛࣗಈੜ AWS ϗϫΠτϖʔύʔ͘Μ
1. ·ͣҰʢֶशෆेͳʣਪ͕ಈ͘͜ͱΛ֬ೝ 2. ͦͷޙվΊͯɺσʔλͷऔಘɾՃ 3. ֶश 4. ৽൛ਪαʔϏεσϓϩΠ σϞͷྲྀΕ
ʮAmazon EC2 isʯΛ༩͑ͨ࣌ͷɺֶशෆेͳਪྫ ਪͷ༷ࢠ ৽͍͠ݴޠ͕ੜ·Ε·ͨ͠ɻ
ʮAmazon EC2 isʯͰɺͦΕͳΓʹֶशͨ͠ޙͷਪྫ ਪͷ༷ࢠ ݴ͍͍ͨ͜ͱΘ͔Γ·ͤΜ͕ɺ୯ޠจ๏ਵ͠·ͨ͠ɻ
γεςϜߏ
γεςϜߏ AWS Batch S3 2.ֶशδϣϒೖ ΤϯδχΞ SpotFleet DeepLearning AMI v2
1.ֶशσʔλೖ 3.δϣϒεέδϡʔϧ 4.σʔλऔಘ & ֶश 5.݁ՌϞσϧΛอଘ ҰൠϢʔβ AWS Lambda 7.Ϟσϧऔಘ & ਪ APIGateway 6.ਪϦΫΤετ 8.݁ՌԠ ECS EC2 g2.2xlarge EC2 g2.2xlarge
ϙΠϯτ
GPU ར༻ AMI Λ͏ • Unmanaged ڥͰ SpotFleet Ͱ҆͘ʂ •
CloudFormation Ͱڥͷల։Λ༰қʹ
NVIDIA-docker, awslogs • NVIDIA-docker ඞਢͰͳ͍ͷͷೖΕΔͱศར • Unmanaged ڥͰ log
CloudWatch Logs ʹ
privileged ϞʔυͰ job Λఆٛ
git ϦϙδτϦ
git clone on your machine! https://github.com/pottava/mxnet-char-lstm
ϋϯζΦϯ͋Γ·͢
COBOL on AWS Batch http://qiita.com/pottava/items/435c65b1fa72cb643f6e
JAWS-UG AI ࢧ෦
ίϯςϯπ • AWS Ͱ AI αʔϏεΛ࣮ɾӡ༻͢ΔͨΊͷ ɹҰൠతͳٕज़ใɺݟɺࣄྫڞ༗ͷ • ͢Ͱʹ׆༻͍ͯ͠Δํ •
ಋೖΛݕ౼͍ͯ͠Δํ • ԿͦΕ͓͍͍͠ͷʁͳํʢ։࠵͝ͱʹқ͕ଟগҧ͍·͢ʣ
ӡӦϝϯόʔ
ࢀߟจݙ ࢀߟจݙ: • AWS Batch – ؆୯ʹ͑ͯޮతͳόονίϯϐϡʔςΟϯάػೳ – AWS https://aws.amazon.com/jp/batch/
• AWS Black Belt Online SeminarʮAWS Batchʯͷࢿྉ͓ΑͼQAެ։ http://aws.typepad.com/sajp/2017/02/aws-black-belt-online-seminar-aws- batch.html#QCPzBdn.twitter_tweet_count_m • re:Invent 2016: AWS Big Data & Machine Learning Sessionsɻ https://aws.amazon.com/blogs/big-data/reinvent-2016-aws-big-data- machine-learning-sessions/