DALL·E: Creating Images from Text

DALL·E[1] is a 12-billion parameter version of GPT-3 trained to generate images from text descriptions, using a dataset of text–image pairs.

DALL·E: Creating Images from Text

Comments

Holotron: A Robotic Exosuit for Virtual Reality | Digital Trends

There was once a time, in the early 1990s to be precise, when the biggest impediment to achieving high-end virtual reality was the fact that the images were too fuzzy and slow-moving. Today, we’ve moved way beyond these early teething challenges. Holotron: A Robotic Exosuit for Virtual Reality | Digital Trends

GPT-J is self-hosted open-source analog of GPT-3: how to run in Docker

Generative Pre-trained Transformer is a generation of models used to produce human-like text based on some initial text (part of dialogs or some task). One of the most "hyped" models is a GPT-3. When you see what GPT-3 generates you feel like "the future is here". GPT-J is self-hosted open-source analog of GPT-3: how to run in Docker

Ironman 2017 - Maastricht

After I had read the Wikipedia page about the Ironman triathlon, my mind went from "insane" to "let's do that before I'm 40". I started running in September 2011, gradually increasing the mileage. In the process switching from regular running shoes to Vibram barefoot shoes. As the big 4-O came closer, it was time to up the game and get serious. Here was not something I was able to accomplish by myself. Exactly two years to the date of the IM I send an email to AVtriathlon, the tri club in my hometown requesting if I could give the running training a whirl. And whirl it was, thanks, Bert, Gert, Wouter and Joop for teaching me to run and having the patience to cope with my very sub-par motor skills. After six months things got better, I could finally do all of the running drills; learned a lot about heart rate, running technique, and pacing. Training for the swim started the same month as the running training. Twice a week, at 6 am, on Mondays and Fridays....

Zeger

Search This Blog