Upgrade to Pro — share decks privately, control downloads, hide ads and more …

[論紹介] Fast Lexically Constrained Decoding with ...

Avatar for onizuka laboratory onizuka laboratory
July 11, 2018
Research
390
0

[論紹介] Fast Lexically Constrained Decoding with Dynamic Beam Allocation for Neural Machine Translation

弊研究室で行なったNAACL読み会の発表資料です。

Avatar for onizuka laboratory

onizuka laboratory

July 11, 2018

More Decks by onizuka laboratory

See All by onizuka laboratory
Phrase-Based & Neural Unsupervised Machine Translation
Avatar for onizuka laboratory onilab
0
120
Tell-and-Answer: Towards Explainable Visual Question Answering using Attributes and Captions
Avatar for onizuka laboratory onilab
0
82
Card-660: A Reliable Evaluation Framework for Rare Word Representation Models
Avatar for onizuka laboratory onilab
0
43
A Word-Complexity Lexicon and A Neural Readability Ranking Model for Lexical Simplification
Avatar for onizuka laboratory onilab
0
140
Integrating Transformer and Paraphrase Rules for Sentence Simplification
Avatar for onizuka laboratory onilab
0
66
An Auto-Encoder Matching Model for Learning Utterance-Level Semantic Dependency in Dialogue Generation
Avatar for onizuka laboratory onilab
0
62
Generating More Interesting Responses in Neural Conversation Models with Distributional Constraints
Avatar for onizuka laboratory onilab
0
110
Modeling Multi-turn Conversation with Deep Utterance Aggregation
Avatar for onizuka laboratory onilab
0
100
Learning Semantic Sentence Embeddings using Pair-wise Discriminator
Avatar for onizuka laboratory onilab
0
130

Other Decks in Research

See All in Research
AIで最適化を解けるか？
Avatar for MIKIO KUBO mickey_kubo
0
100
Using our influence and power for patient safety
Avatar for Helen Bevan helenbevan
0
360
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
Avatar for Shunsuke KITADA shunk031
0
120
非試合日の野球場を楽しむためのARホームランボールキャッチ体験システムの開発 / EC79-miyazaki
Avatar for yumulab yumulab
0
210
AI Agentの精度改善に見るML開発との共通点 / commonalities in accuracy improvements in agentic era
Avatar for shimacos shimacos
6
1.7k
セマンティック通信勉強会 6Gに向けたデバイス間効率的な通信の技術紹介・課題・今後展望
Avatar for SatAI.challenge satai
3
150
LLM の Attention 機構まとめ — 数式・計算量・メモリ
Avatar for プワー puwaer
7
2k
討議：RACDA設立30周年記念都市交通フォーラム2026
Avatar for Traffic Brain trafficbrain
0
940
Claude Code × autoresearch 実践
Avatar for 数理の弾丸 mathbullet
0
150
進学校の生徒にはア行の苗字が多いのか
Avatar for Takashi Ozeki ozekinote
0
430
IEEE AIxVR 2026 Keynote Talk: "Beyond Visibility: Understanding Scenes and Humans under Challenging Conditions with Diverse Sensing"
Avatar for Mariko Isogawa miso2024
0
200
Ankylosing Spondylitis
Avatar for Charles Holtzkampf ankh2054
0
170

Featured

See All Featured
世界の人気アプリ100個を分析して見えたペイウォール設計の心得
Avatar for Akihiro Kokubo akihiro_kokubo
PRO
71
40k
Impact Scores and Hybrid Strategies: The future of link building
Avatar for Tamara Novitovic tamaranovitovic
0
300
職位にかかわらず全員がリーダーシップを発揮するチーム作り / Building a team where everyone can demonstrate leadership regardless of position
Avatar for MADOX madoxten
62
54k
Ecommerce SEO: The Keys for Success Now & Beyond - #SERPConf2024
Avatar for Aleyda Solis aleyda
1
2k
Skip the Path - Find Your Career Trail
Avatar for Mark Kilby mkilby
1
140
Highjacked: Video Game Concept Design
Avatar for Ryan Kendrick rkendrick25
PRO
1
380
The innovator’s Mindset - 
Leading Through an Era of Exponential Change - McGill University 2025
Avatar for Jean-Marc De Jonghe jdejongh
PRO
1
190
Improving Core Web Vitals using Speculation Rules API
Avatar for Sergey Chernyshev sergeychernyshev
21
1.5k
Designing Experiences People Love
Avatar for Jonathan Moore moore
143
24k
Max Prin - Stacking Signals: How International SEO Comes Together (And Falls Apart)
Avatar for Tech SEO Connect techseoconnect
PRO
0
180
Have SEOs Ruined the Internet? - User Awareness of SEO in 2025
Avatar for Akash Hashmi akashhashmi
0
360
Raft: Consensus for Rubyists
Avatar for Patrick Van Stee vanstee
141
7.5k

Transcript

  1. 'BTU-FYJDBMMZ$POTUSBJOFE%FDPEJOH XJUI%ZOBNJD#FBN"MMPDBUJPO GPS/FVSBM.BDIJOF5SBOTMBUJPO .BUU1PTUBOE%BWJE7JMBS "NB[PO3FTFBSDI #FSMJO (FSNBOZ 1SPDPG/""$- QQ 1SFTFOUFECZ5PNPZVLJ,BKJXBSB

  2. $POUSPMMJOH/.5PVUQVU • /.5SFNPWFTNBOZXBZTPGNBOVBMMZHVJEJOH UIFUSBOTMBUJPOQSPDFTTJOPMEFSQBSBEJHNT 1#4.5  • *ODPSQPSBUFEPNBJOTQFDJGJDEJDUJPOBSJFT • 'PSDFBUSBOTMBUJPODIPJDFGPSDFSUBJOXPSET

    • -FYJDBMMZ$POTUSBJOFE%FDPEJOHGPS4FRVFODF(FOFSBUJPO 6TJOH(SJE#FBN4FBSDI )PLBNQ BOE-JV "$- • 6TFSTDBOTQFDJGZUIFXPSETUIBUNVTUBQQFBSJOUIFPVUQVU • 6TFGVMGPSUBTLTTVDIBTQPTUFEJUJOH DBQUJPOHFOFSBUJPO   /JFNBOE IBUEJF"CTJDIU FJOF .BVFS[V CBVFO /JFNBOE IBUEJF"CTJDIU FJOF .BVFS[V FSSJDIUFO ,FJOFS IBUEJF"CTJDIU FJOF .BVFS[V CBVFO FSSJDIUFO LFJOFS -FYJDBMMZ$POTUSBJOFE%FDPEJOH /PPOFIBTUIFJOUFOUJPO PGCVJMEJOHBXBMM

  3. (#4IBTBMBSHF3VOUJNF$PNQMFYJUZ %FDPEJOH$PNQMFYJUZ 0 /L$ (SJE#FBN4FBSDI )PLBNQ BOE-JV  0 /L

    %ZOBNJD#FBN"MMPDBUJPO 5IJT8PSL /TFOUFODFMFOHUI LCFBNTJ[F $OVNCFSPGDPOTUSBJOUT (#4IBTUXPQSPCMFNT • %FDPEJOHDPNQMFYJUZJTMJOFBSJODPOTUSBJOUT • $BOOPUTQFDJGZUIFCFBNTJ[FBUNPEFMMPBEUJNF 

  4. %#"%ZOBNJD#FBN"MMPDBUJPO • 4JNQMZSFQMBDFUIF,#&45 JNQMFNFOUBUJPO GSPNUIFBMHPSJUINPGTUBOEBSECFBNEFDPEJOH • *OTUFBEPGTFMFDUJOHUIFUPQLJUFNT  (FOFSBUJOHBMJTUPGDBOEJEBUFT 8FOPXOFFEUPFOTVSFUIBUDBOEJEBUFTQSPHSFTT

    UISPVHIUIFTFUPGQSPWJEFEDPOTUSBJOUT  "MMPDBUJOHUIFCFBNBDSPTTUIFDPOTUSBJOUCBOLT 8JUIBGJYFETJ[FECFBNBOEBSCJUSBSZOVNCFSPG DPOTUSBJOUT XFOFFEUPGJOEBOBMMPDBUJPOTUSBUFHZ  GVODUJPO #&".4&"3$) / L CFBNˡ %&$0%&3*/*5 L GPS UJNFTUFQUJO/EP TDPSFT%&$0%&345&1 CFBN CFBNˡ ,#&45 TDPSFT SFUVSO CFBN<>

  5. 4UFQ(FOFSBUJOHUIFDBOEJEBUFTFU $BOEJEBUFTFU ˒ LCFTUJUFNT JF OPSNBMUPQL ˖ TJOHMFCFTUUPLFO ˠ VOGJMMFEDPOTUSBJOUT

     #FBNTJ[F L  7PDBCVMBSZTJ[F c7c  $POTUSBJOUT $  XPSET QISBTF

  6. 4UFQ"MMPDBUJOHUIFCFBN "MMPDBUFTJ[FLCFBNBDSPTT$ DPOTUSBJOUCBOLT #BOL 5IFQPSUJPOPGUIFCFBNSFTFSWFEGPSJUFNT IBWJOHNFUUIFTBNFOVNCFSPGDPOTUSBJOUT "MMPDBUJPOTUSBUFHZTFUUJOHFBDICJOTJ[FUP⌊L$⌋ "OZSFNBJOJOHTMPUTBSFBTTJHOFEUPUIFNPTUDPOTUSBJOFECBOL 

  7. &YQFSJNFOUBM4FUVQ • %BUB • 5SBJO8.5&OHMJTI(FSNBOUSBJOJOHDPSQPSB • %FW5FTUOFXTUFTU  • 1SFQSPDFTT

    • .PTFTUPLFOJ[FS • +PJOU#1&WPDBCVMBSZXJUILNFSHFPQFSBUJPO • .PEFM • MBZFS3//XJUIBUUFOUJPO • "EBNPQUJNJ[FSXJUIBCBUDITJ[FPG • &BSMZTUPQQJOH DSPTTFOUSPQZ • #BTFMJOFT • (SJE#FBN4FBSDI LC $  C   

  8. 3VOOJOHUJNF MPXFSJTCFUUFS BOE#-&6TDPSF • 5IFMJOFBSUSFOEJO$ DPOTUSBJOUT JTDMFBSGPS(#4 • 5IFDPOTUBOUUSFOEJO$ JTDMFBSGPS%#"

    • %#"JTBCPVUYTMPXFSUIBOVODPOTUSBJOFEEFDPEJOH  #-&6 #-&6 #-&6

  9. $POTUSBJOFEXPSETBSFQMBDFEJOUIFDPSSFDUQPTJUJPO 1FBSTPO`TS  

  10. $POTUSBJOFEXPSETBSFQMBDFEJOUIFDPSSFDUQPTJUJPO 

  11. .PEFMTDPSFHPFTEPXOCVU#-&6TDPSFHPFTVQ 5IJTJTBOPCTFSWBUJPOSFMBUFEUPBSFQPSU UIBU #-&6XPSTFOTJGUIFCFBNTJ[FJTNBEFUPPMBSHF  ,PFIOBOE,OPXMFT 8/.5 4JY$IBMMFOHFTGPS/FVSBM.BDIJOF5SBOTMBUJPO

  12. 'BTU-FYJDBMMZ$POTUSBJOFE%FDPEJOHXJUI%#"GPS/.5 • 5IJTTUVEZNBEFJUQPTTJCMFUPDPOUSPM/.5PVUQVU VTJOH-FYJDBMMZ$POTUSBJOFE%FDPEJOH • *O(#4 UIFDPNQVUBUJPOBMDPTUJODSFBTFEMJOFBSMZ XJUI DPOTUSBJOUT CVUJO%#"

    JUEFDSFBTFEUPDPOTUBOU PSEFS • $POTUSBJOUTBSFDPSSFDUMZQMBDFEXPBMJHONFOUJOGPSNBUJPO