版權說明:本文檔由用戶提供并上傳,收益歸屬內容提供方,若內容存在侵權,請進行舉報或認領
文檔簡介
?
2023,
Amazon
Web
Services,
Inc.
or
its
affiliates.?
2023,
Amazon
Web
Services,
Inc.
or
its
affiliates.Generative
AIon
Amazon
SageMakerBenoit
de
PatoulAI/ML
Specialist
Solutions
Architect?
2023,
Amazon
Web
Services,
Inc.
or
its
affiliates.Wework
backwards
from
our
customers
and
wevalue
your
feedback!?
20232,
Amazon
Web
Services,
Inc.
or
its
affiliates.What
is
Generative
AI3Generative
AI
is
the
fastest
growing
trend
in
AI4Days
since
repository
was
startedGithubStars400800
1,200
1,600
2,000
2,4002,8003,20040,000Stable
DiffusionEthereumBitcoinDeveloper
adoptionStable
Diffusion
accumulated
40k
starson
GitHub
in
its
first
90
daysConsumer
adoptionChatGPT
reachedthe
1
millionusers
mark
in
just
5
days1M500kChatGPT5
daysInstagram75
daysSpotify150
days0255012515075
100Days
since
launchStylized
pathWhat
is
generative
AI?Reduces
timeand
costtodevelop
ML
models
andinnovate
fasterApplicable
to
many
use
caseslike
text
summarization,question
answering,
digitalartcreation,
code
generation,
etc.Powered
by
foundation
modelspre-trained
on
large
sets
ofdatawith
several
hundredbillionparametersTasks
canbe
customized
forspecific
domains
withminimal
fine-tuningAI
that
can
produce
originalcontent
closeenoughtohuman
generated
contentfor
real-world
tasksGenerative
AI
is
emerging
across
a
range
of
domains
…ProductivityText
generationChatVirtual
assistantSummarizationText
extractionSearchCodegenerationImagegenerationImageclassificationCreatemusicGeneratevideos6Project
yProject
xProject
xWhat
is
a
Foundation
ModelPreprocessingTask-specificModel
TrainingTask-specificDeploymentModelValidation
/TestingTask-specificDeploymentConventionalML
flowMLflow
withFoundationModelsHyper-parameterTuning…PreprocessingFoundationModel
Training/
TuningTask-specificModel(s)FoundationModelFinetuningTask-specificDeploymentDataDataIteration(s)7?
20232,
Amazon
Web
Services,
Inc.
or
its
affiliates.Generative
AI
in
AWS12?
2023,
Amazon
Web
Services,
Inc.
or
its
affiliates.Option
1:Foundation
modelsavailable
throughSageMakerJumpStart13You
can
try
out
models
for
free..!Machine
Learning
Hub
for
SageMakerBrowse
through
~400
contents
including,
built-in
algorithms
with
pre-trainedmodels,
(New)
Foundation
Models,
solution
templates,
and
example
notebooksPre-built
training
and
inference
scriptsCompatible
with
SageMaker
and
configurable
with
custom
datasetUI
as
well
as
API
based
machine
learningUse
UI
for
single
click
model
deployment
or
API
for
Python
SDK
based
workflowNotebook
with
examplesJumpStart
lets
you
jump
into
notebook
to
use
selected
model
with
examplestoguide
customers
through
entire
ML
workflow(New)
Share
and
collaborate
within
an
organizationShare
models
and
notebookswith
others
within
your
organization,
and
allow
themto
trainwiththeir
own
data
or
deploy
as-isforinferencingSageMakerJumpStartEasilyaccess
ML
assets
andquickly
bring
MLapplications
tomarketFoundation
models
with
SageMaker
JumpStart:How
it
worksAmazon
SageMakerJumpStartAccess
and
try
out
public
andproprietary
FMs,
and
customizeand
integrate
them
into
yourgenerative
AI
applicationsDeployDeploy
the
modeland
run
inferencefor
your
generativeAI
use
caseBrowseBrowse
publicand
proprietary
FMsExperimentExperiment
with
FMsbefore
choosing
amodelfor
deploymentCustomizeCustomize
selectedFM
with
your
owndataset
withouttraining
from
scratch15Foundation
models
available
on
SageMaker
JumpStartfor
self-managed
accessModelsCoheregenerate-medTasksText
generationInformationextractionQuestionansweringSummarizationModelsText2ImageUpscalingTasksGeneratephoto-realisticimages
fromtext
inputImprove
qualityof
generatedimagesFeaturesFine-tuning
onSD
2.1
modelModelsFlan
T-5
models(8
variants)DistilGPT2,
GPT2Bloom
models(3
variants)TasksMachinetranslationQuestionansweringSummarizationAnnotation
Data
generationModelsAlexaTM20BTasksMachinetranslationQuestionansweringSummarizationAnnotation
Data
generationModelsJurassic-1Grande
17BTasksText
generationLong-formgenerationSummarizationParaphrasingChatInformationextractionQuestionansweringClassificationModelsLyra-Fr10BTasksText
generationKeywordextractionInformationextractionQuestionansweringSummarizationSentimentanalysisClassificationPublicly
available Proprietary
models163
ways
to
use
foundation
models
with
SageMakerJumpStartSageMaker
StudioOne-step
deploySageMakerNotebooks17AWS
ManagementConsole
PreviewOption
2
-
Amazon
BedrockThe
easiest
way
to
build
andscale
generative
AIapplications
with
FMsNEW18?
2023,
Amazon
Web
Services,
Inc.
or
its
affiliates.Amazon
Bedrock
key
benefitsEnhance
your
dataprotection
usingcomprehensive
AWSsecurity
capabilitiesAcceleratedevelopment
ofgenerative
AIapplications
usingFMs
through
an
API,without
managinginfrastructure19ChooseFMs
fromAI21
Labs,
Anthropic,Stability
AI,
andAmazon
to
find
theright
FM
for
your
usecasePrivately
customizeFMsusing
yourorganization’s
dataUse
AWS
tools
andcapabilities
that
youare
familiar
with
todeploy
scalable,reliable,
and
securegenerative
AIapplicationsBedrock
supports
a
wide
range
of
foundation
modelsTitan
TextTitanEmbeddingsFMs
from
AmazonFMsfrom
AI21
Labs,
Anthropic,and
Stability
AIJurassic-2ClaudeStabl
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網頁內容里面會有圖紙預覽,若沒有圖紙預覽就沒有圖紙。
- 4. 未經權益所有人同意不得將文件中的內容挪作商業(yè)或盈利用途。
- 5. 人人文庫網僅提供信息存儲空間,僅對用戶上傳內容的表現方式做保護處理,對用戶上傳分享的文檔內容本身不做任何修改或編輯,并不能對任何下載內容負責。
- 6. 下載文件中如有侵權或不適當內容,請與我們聯(lián)系,我們立即糾正。
- 7. 本站不保證下載資源的準確性、安全性和完整性, 同時也不承擔用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。
最新文檔
- 家庭教育中的信息素養(yǎng)教育與孩子未來發(fā)展競爭力
- 有關共同富裕的課程設計
- 二零二五年度建材行業(yè)標準制定居間服務合同2篇
- 美業(yè)營銷運營課程設計
- 智能交通廣告投放服務合同
- 中國商業(yè)新業(yè)態(tài)的宏觀經濟發(fā)展效應分析
- 孩子情感教育的多元化方法與實踐
- 文化傳媒項目合作框架協(xié)議
- 機械設計課程設計浙大
- 二零二五年度建筑工程班組進度管理勞務服務合同3篇
- 課程設計YA32-350型四柱萬能液壓機液壓系統(tǒng)設計
- 圍堰高噴防滲墻工程監(jiān)理實施細則
- (精心整理)系動詞練習題
- 婚戀觀教育--保持軍人本色樹立正確婚戀觀
- 體彩排列五歷史數據
- 中國工業(yè)數據庫介紹
- 弱電智能化設計服務建議書(共35頁)
- 中國銀監(jiān)會關于規(guī)范中長期貸款還款方式的通知
- 通信工程外文文獻(共12頁)
- 汽車底盤維修實訓考核表(共24頁)
- 煉鐵廠3#燒結主抽風機拆除安全專項方案
評論
0/150
提交評論