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
LLM とプロンプトエンジニアリング/チューターをビルドする / LLM and Prompt...
Search
Kenji Saito
PRO
April 15, 2024
Technology
0
350
LLM とプロンプトエンジニアリング/チューターをビルドする / LLM and Prompt Engineering and Building Tutors
早稲田大学大学院経営管理研究科「プロンプトエンジニアリング ─ 生成AIの応用」2024 春の第1-2回で使用したスライドです。
Kenji Saito
PRO
April 15, 2024
Tweet
Share
More Decks by Kenji Saito
See All by Kenji Saito
市民科学にAI はどう活用できるか / How AI Can Be Used for Citizen Science
ks91
PRO
0
11
グリーンマイニングが Bitcoin プロトコルに及ぼす影響 / Impact of Green Mining on the Bitcoin Protocol
ks91
PRO
0
18
FinTech 13-14 : Ideathon, Presentations and Conclusions
ks91
PRO
0
55
デザイン相談会 / Design Consultation
ks91
PRO
0
13
FinTech 11-12 : Cyber-Physical Society and Future of Finance
ks91
PRO
0
52
メタ自然選択と製品トレーサビリティー / Meta-Natural Selection and Product Traceability
ks91
PRO
0
8
伝統的金融に呑まれる分散型金融 / Decentralised Finance Engulfed by Traditional Finance
ks91
PRO
0
16
ウェブサービスデザイン 2 / Web Service Design 2
ks91
PRO
0
26
生成AI による論文執筆サポートの手引き(ワークショップ) / A guide to supporting dissertation writing with generative AI (workshop)
ks91
PRO
0
250
Other Decks in Technology
See All in Technology
New Relicを活用したSREの最初のステップ / NRUG OKINAWA VOL.3
isaoshimizu
3
660
アジャイルでの品質の進化 Agile in Motion vol.1/20241118 Hiroyuki Sato
shift_evolve
0
190
LINEヤフーにおけるPrerender技術の導入とその効果
narirou
1
240
RubyのWebアプリケーションを50倍速くする方法 / How to Make a Ruby Web Application 50 Times Faster
hogelog
3
950
Making your applications cross-environment - OSCG 2024 NA
salaboy
0
210
Adopting Jetpack Compose in Your Existing Project - GDG DevFest Bangkok 2024
akexorcist
0
120
CDCL による厳密解法を採用した MILP ソルバー
imai448
3
200
rootlessコンテナのすゝめ - 研究室サーバーでもできる安全なコンテナ管理
kitsuya0828
3
390
OCI 運用監視サービス 概要
oracle4engineer
PRO
0
4.8k
TypeScriptの次なる大進化なるか!? 条件型を返り値とする関数の型推論
uhyo
2
1.8k
20241120_JAWS_東京_ランチタイムLT#17_AWS認定全冠の先へ
tsumita
2
320
Flutterによる 効率的なAndroid・iOS・Webアプリケーション開発の事例
recruitengineers
PRO
0
130
Featured
See All Featured
Building a Modern Day E-commerce SEO Strategy
aleyda
38
6.9k
"I'm Feeling Lucky" - Building Great Search Experiences for Today's Users (#IAC19)
danielanewman
226
22k
What’s in a name? Adding method to the madness
productmarketing
PRO
22
3.1k
Reflections from 52 weeks, 52 projects
jeffersonlam
346
20k
Designing for humans not robots
tammielis
250
25k
Save Time (by Creating Custom Rails Generators)
garrettdimon
PRO
27
840
GitHub's CSS Performance
jonrohan
1030
460k
Thoughts on Productivity
jonyablonski
67
4.3k
Into the Great Unknown - MozCon
thekraken
32
1.5k
[RailsConf 2023 Opening Keynote] The Magic of Rails
eileencodes
28
9.1k
Code Review Best Practice
trishagee
64
17k
Java REST API Framework Comparison - PWX 2021
mraible
PRO
28
8.2k
Transcript
Generative, Volatile, Uncertain, Complex and Ambiguous — generated by Stable
Diffusion XL v1.0 — AI 2024 1-2 LLM (WBS) 2024 1-2 LLM — 2024-04-15 – p.1/37
https://speakerdeck.com/ks91/collections/prompt-engineering-2024-spring 2024 1-2 LLM — 2024-04-15 – p.2/37
( ) ( ) ( ) CSO (Chief Science Officer)
1993 ( ) 2006 ( ) SFC 23 P2P (Peer-to-Peer) 2011 ( ) 2018 2019 VR 2021.9 & VR 2022.3 2023 AI VR&RPG 2023.5 “Don’t Be So Serious” 2023 2024 AI( ) 2024 “ALOHA FROM HAWAII” → ( ) 2024 1-2 LLM — 2024-04-15 – p.3/37
Dropbox Dropbox ( ) ( ) ( ) 2024 1-2
LLM — 2024-04-15 – p.4/37
LLM ( ) GPTs 2024 1-2 LLM — 2024-04-15 –
p.5/37
(LLM : Large Language Model) AI (AGI : Artificial General
Intelligence) ( . . . ^^;) 2024 1-2 LLM — 2024-04-15 – p.6/37
( ) LLM LLM LLM ⇒ 2024 1-2 LLM —
2024-04-15 – p.7/37
( ) 1 4 15 LLM • 2 4 15
• 3 4 22 I 4 4 22 5 4 29 6 4 29 7 5 13 8 5 13 9 5 20 II 10 5 20 (1) 11 5 27 (2) 12 5 27 (3) 13 6 3 & 14 6 3 & W-IOI 2024 1-2 LLM — 2024-04-15 – p.8/37
( 20 ) 1 • 2 • 3 Discord &
• 4 5 6 RPG 7 “September 12th” 8 9 10 ∼ 11 Linux (Windows )(Mac ) 12 Open Interpreter ∼ 13 14 AGI (Artificial General Intelligence) 7 (5/6 ) / (2 ) OK / 2024 1-2 LLM — 2024-04-15 – p.9/37
. . . . . . ( ) ( 20
×14 ) 2024 1-2 LLM — 2024-04-15 – p.10/37
(2 )(160 ) (10∼20 ) and/or I (60∼80 ) II
(60∼80 ) 2024 1-2 LLM — 2024-04-15 – p.11/37
Moodle ( Q&A ) ( ) Discord ( ) ←
( ) 2024 1-2 LLM — 2024-04-15 – p.12/37
( ) 2024 1-2 LLM — 2024-04-15 – p.13/37
OpenAI GPT-4 (Generative Pretrained Transformer 4) API (Application Programming Interface)
Discord ( ) ( AIC ) /chat GPT-4 GPT ChatGPT (ChatGPT ) Discord GPT-4 GPT-4 ChatGPT Plus 2024 1-2 LLM — 2024-04-15 – p.14/37
Discord web 80 ( ) ( ) 2024 1-2 LLM
— 2024-04-15 – p.15/37
LLM ( ) . . . ( ) 2024 1-2
LLM — 2024-04-15 – p.16/37
B. , , 1968 = ⇒ ( ) ( )
2024 1-2 LLM — 2024-04-15 – p.17/37
: ⇒ = 30 . . . ( ) 30
⇒ + LLM (2023 3 ) + LLM ( ) 2024 1-2 LLM — 2024-04-15 – p.18/37
m P(w1 , . . . , w m )
(Wikipedia) 1 (Wikipedia) : ( ) ← (Generative Pre-training) : ( ) ( ) https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/ 2024 1-2 LLM — 2024-04-15 – p.19/37
ChatGPT GPT ChatGPT GPT OpenAI (GPT-3.5, GPT-4) GPT Generative Pre-trained
Transformer ( ) GPT ( ) GPT-3.5, GPT-4 2024 1-2 LLM — 2024-04-15 – p.20/37
GPT 2024 1-2 LLM — 2024-04-15 – p.21/37
ChatGPT ChatGPT ← LLM LLM → 200 ( ) 2024
1-2 LLM — 2024-04-15 – p.22/37
GPT 2024 1-2 LLM — 2024-04-15 – p.23/37
( ) GPT 3 1. (GPT ) 2. ( )
3. ( ) 3 (↑ 3. ) 3. (GPT ) 2024 1-2 LLM — 2024-04-15 – p.24/37
GPTs ( ) 2024 1-2 LLM — 2024-04-15 – p.25/37
GPTs ʂ (GPT) ChatGPT 2024 1-2 LLM — 2024-04-15 –
p.26/37
GPT Ζ͏ͥ + GPT 2024 1-2 LLM — 2024-04-15 –
p.27/37
ͬͪ͜ͰϏϧμʔͱର͠ͳ͕Β ৽͍͠ (15 ΛΕ·͢ ͬͪ͜Ͱ৽͍͠ (15 Λࢼͤ·͢ ͲΜͳϓϩϯϓτʹͳ͍ͬͯΔ͔ΛݟͨΓ ࡉ͔͘मਖ਼ͨ͠Γઃఆͨ͠ΓͰ͖·͢ GPT
2024 1-2 LLM — 2024-04-15 – p.28/37
( ) GPT 3 1. (GPT ) 2. ( )
3. ( ) 3 (↑ 3. ) 3. (GPT ) 2024 1-2 LLM — 2024-04-15 – p.29/37
NEO NEO (CC-BY-SA 3.0) ( ) 2024 1-2 LLM —
2024-04-15 – p.30/37
( ) Ϗϧμʔͱ͢ εϨου ૬ஊ͢Δ ର͢Δ ͪΐͬͱ ҧ͏ͳ ςΩετνϟϯωϧ ͔Β
%JTDPSE͔Βίϐʔͨ͠ ༰ΛϝϞாͰ ฤूͯ͠ ίϐϖ͢Δͱ Α͍͔ ˣ ߋ৽εϨου͔Β ςΩετνϟϯωϧ ͔Β ςΩετνϟϯωϧ ͔Β Α͠ࢼͯ͠ΈΑ͏ ॳճDIBU DIBU ॳճ CVJME ߋ৽ VQEBUF આ໌ ࢦࣔΛ ೖྗ͢Δ ΞγελϯτΛ࡞Γ ߋ৽͢ΔεϨου ΞγελϯτΛ ࢼ͢εϨου 2024 1-2 LLM — 2024-04-15 – p.31/37
4 Discord Discord & 2024 1-2 LLM — 2024-04-15 –
p.32/37
GPT-Discord Bot System ϝϯόʔ͕ॴଐ͢Δ৫ ϝϯόʔ େنݴޠϞσϧ (155VSCP ࢲͨͪͷ νʔϜ͕։ൃ εϨου
ࢲ ࢲͨ ͨͪ ͪ νϟοταʔϏε %JTDPSE ϫʔΫεϖʔε %JTDPSE αʔόʔ PE νϟοτϘοτ HQUEJTDPSECPU ৗ࣌Քಇɾৗ࣌ଓͷ ԿΒ͔ͷϚγϯ ƅ˜ƅ ů ։࢝ Ξγελϯτ "1* εϥογϡίϚϯυ ᶃ ᶄ ᶅ ᶆ ᶇ ίϚϯυʹΑΔ੍ޚ Ξγελϯτͱͯ͠ର ಡΈग़͠ɾॻ͖ࠐΈ ର ର ର ର ʲΞγελϯτ "1*ʳ ɾΞγελϯτͷੜ $ ɾऔಘ 3 ɾߋ৽ 6 ɾআ % ɾ 0QFO"* εϨουͷੜɾॻ͖ࠐΈ ϝϯόʔɺίϚϯυΛ௨ͯ͠ ʮϏϧμʔΞγελϯτʯͱର͠ͳ͕Β νϡʔλʔ Ξγελϯτ Λੜͨ͠Γɺ ੜࡁͷνϡʔλʔΛमਖ਼ͨ͠Γɺ νϡʔλʔΛݺͼग़ͯ͠ରͰ͖·͢ɻ ʲίϚϯυҰཡʳ CVJME DIBU EFMFUF MJTU TIPX VQEBUF ʜΞγελϯτΛੜ ʜΞγελϯτͱର։࢝ ʜΞγελϯτΛআ ʜΞγελϯτҰཡ ʜΞγελϯτΛදࣔ ʜΞγελϯτΛमਖ਼ GPTs × ⇒ 2024 1-2 LLM — 2024-04-15 – p.33/37
code interpreter FILES ( OK) ( ) 2024 1-2 LLM
— 2024-04-15 – p.34/37
2024 1-2 LLM — 2024-04-15 – p.35/37
1. (1) AI (2) 2024 4 18 ( ) 23:59
JST ( ) Waseda Moodle 2024 1-2 LLM — 2024-04-15 – p.36/37
2024 1-2 LLM — 2024-04-15 – p.37/37