Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
Searching for Needles in a Haystack: On the Rol...
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
Hiroyuki Deguchi
October 03, 2023
Research
180
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
Searching for Needles in a Haystack: On the Role of Incidental Bilingualism in PaLM’s Translation Capability
Hiroyuki Deguchi
October 03, 2023
More Decks by Hiroyuki Deguchi
See All by Hiroyuki Deguchi
20260427 Ehime DS Seminar: MBR-CBDT Decoding
de9uch1
1
53
20260624 NLP colloquium: 単一のhubテキストがCLIPを壊す:hubnessによる埋め込みの脆弱性特定
de9uch1
2
280
20250226 NLP colloquium: "SoftMatcha: 10億単語規模コーパス検索のための柔らかくも高速なパターンマッチャー"
de9uch1
1
850
20240820: Minimum Bayes Risk Decoding for High-Quality Text Generation Beyond High-Probability Text
de9uch1
0
380
サブセット探索を用いた高速なkNNニューラル機械翻訳
de9uch1
0
180
20240226_AAMT-Japio
de9uch1
0
210
Paper Reading: Sampling-Based Approximations to Minimum Bayes Risk Decoding for Neural Machine Translation
de9uch1
0
240
My Research Environmental Setup
de9uch1
0
350
Nearest Neighbor Machine Translation
de9uch1
0
300
Other Decks in Research
See All in Research
[IR Reading 2026春 論文紹介] LLM-based Listwise Reranking under the Effect of Positional Bias (ECIR 2026) /IR-Reading-2026-Spring
koheishinden
PRO
0
450
量子サマースクール2026「量子計算機アーキテクチャ分野の概観」
youten622
1
950
Karkada さんの論文 × 2 の紹介: (1) Closed-Form Training Dynamics Reveal Learned Features and Linear Structure in Word2Vec-like Models, (2) Symmetry in language statistics shapes the geometry of model representations
eumesy
PRO
1
800
PHTalks Bengaluru - SSRF When All Else Fails
dk999
0
1.2k
敵対生成プロンプト同時探索による内省型プロンプト最適化
kinoue_smarthr
0
430
Pretrain Where? Investigating How Pretraining Data Diversity Impacts Geospatial Foundation Model Performance
satai
3
160
Language and AI
ayaniwa
0
310
Sleuthcon Keynote - How Cybercriminals (ab)use AI
fr0gger
0
350
視覚若手の会LENSって何??
mickey_0226
0
320
Visual SLAM未来予測 / Future Prediction in Visual SLAM
koide3
1
1.1k
Anthropic が提案する LLM の内部状態を自然言語で説明可能にした Natural Language Autoencoders / Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations
shunk031
0
320
進化?迷走?CasualConc ファミリーアプリの現在地 @ 英語コーパス学会 2026
casualconc
0
120
Featured
See All Featured
How to optimise 3,500 product descriptions for ecommerce in one day using ChatGPT
katarinadahlin
PRO
3
3.8k
Amusing Abliteration
ianozsvald
1
320
What does AI have to do with Human Rights?
axbom
PRO
1
2.4k
Creating an realtime collaboration tool: Agile Flush - .NET Oxford
marcduiker
35
2.6k
Music & Morning Musume
bryan
48
7.4k
Refactoring Trust on Your Teams (GOTO; Chicago 2020)
rmw
35
3.8k
Save Time (by Creating Custom Rails Generators)
garrettdimon
PRO
33
5k
Bridging the Design Gap: How Collaborative Modelling removes blockers to flow between stakeholders and teams @FastFlow conf
baasie
0
710
Bootstrapping a Software Product
garrettdimon
PRO
306
120k
Visual Storytelling: How to be a Superhuman Communicator
reverentgeek
2
700
Mozcon NYC 2025: Stop Losing SEO Traffic
samtorres
1
570
How to train your dragon (web standard)
notwaldorf
97
6.8k
Transcript
None
◼ [1] ◼ ◼ ◼ [1] PaLM: Scaling Language Modeling
with Pathways (Chowdhery+, JMLR vol.24, no.240, 2023)
[1] ◼ ◼ ⚫ ⚫ ⚫ ⚫ ◼ ◼ [1]
PaLM: Scaling Language Modeling with Pathways (Chowdhery+, JMLR vol.24, no.240, 2023)
◼ ⚫ ⚫ ◼ ⚫ ⚫ ▶ ▶
◼ ⚫ ⚫ [2] ⚫ [2] ◼ ⚫ ⚫ [2]
Language-agnostic BERT Sentence Embedding (Feng+, ACL2022)
◼ ⚫ ⚫
◼ ⚫ ◼ ◼
◼ ◼ ◼
◼ ⚫ ▶ ▶ ⚫ ⚫ ◼ ⚫ ▶
◼ ⚫ ⚫ ⚫ ▶ ▶ ▶
◼ ⚫ ⚫ ▶ ▶
◼ ⚫ ▶ ▶ ◼ ⚫ ⚫ ▶ ⚫