人工知能(無能?)カラオケ!! – 畳み込みニューラルネットワークによる動画の情景解析に基づく歌詞の自動生成 https://t.co/ltVOBHqPhu  2016年に余興で作ったシステム。CLIPとかGPT-3で今本気で同じものを作ったどうなるかな... とか妄想中

1 5



📋#CryptoMERZ: driven

🖼️ M006: Aitropia — The Cloud Fortress

🔗 https://t.co/EsFssr432c

🌊: https://t.co/lLhqCCnow8

🗝️ Poem (GPT-3), original/animated image



🔭

5 17

https://t.co/uN8hHApNKg
OpenAIが開発したTransformerベースのtext-to-image(文書から画像を生成)モデル。論文は未発表だが120億のパラメータ(GPT-3の1/10程度)をもつ。単一の物体だけでなく複数の物体も描写可能で、存在しない概念(アボカド椅子)も上手く描写できる。

11 62

Open AI, GPT-3 기술, 텍스트로 이미지 제작도 지원하는군요

첨부 이미지는 "아보카도 모양의 안락의자" 로 생성한 이미지

DALL·E: Creating Images from Text https://t.co/KnkvVtIzKC

4 2

Rather than collect 1000s of supervillain bios, train a neural net on them, and style-transfer them all to be about AI, I just had GPT-3 generate the resulting blog post.

Then I collected my favorite AI supervillains and illustrated them.
https://t.co/I5vtnj6eih

79 297

In this week's letter 💌:

😮 Figma designing for you
🤖 GPT-3 design hype
💫 Design sprint as it is

Didn't get the message? Join the band 👉 https://t.co/yBsph1OViO
In the meantime, we've summarised the GPT-3 hype here 👀 https://t.co/3vffb2mbwJ

1 3

hello GPT-3 "draw the perfect horse"

0 4