Dmytro Kalpakchi

21 Followers
10 Following
45 Posts

PhD @ KTH Royal Institute of Technology with interests in Natural Language Processing and Generation

Author and maintainer of UDon2 (https://github.com/udon2/udon2) and Textinator (https://github.com/dkalpakchi/Textinator)

Opinions are my own and represent neither the official position of my current nor my previous employers.

Personal websitehttps://dkalpakchi.github.io/
Githubhttps://github.com/dkalpakchi/
My previous screenshot example of #ImageGen was a bit hard requiring to generate a real question. Although I got neither a question, nor any text in English last time, I thought let's go easier now and prompt with 'a T-shirt with a print "a fan of Stockholm"' #dalle2
stabilityai/stable-diffusion-2 · Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

Finally, I got a tip to use Midjourney, which got fed again exactly the same prompt. The result was this 👇

#NLProc #midjourney

Then I was advised to try Stable diffusion, which I did with exactly the same prompt and I got this👇

#NLProc #StableDiffusion

OK, so I've tried DALLE-2 for the first time and wanted it to generate "a screenshot of a system capable of generating multiple choice questions for reading comprehension". This is what I got, which is good only if you look from afar...

#NLProc #dalle2