Talking counterfactual #6: ChatGPT
August 12, 2023
The pace of development in the field of large language models has been staggering throughout the last months. Just when I settled on a new candidate for “Talking counterfactual” a new model with supposedly superior capabilities was released, be it LLaMA (now already LLaMA 2), MPT, Falcon, Cohere’s models, Claude,...
Read moreDo we dream like Dall-E?
Dezember 12, 2022
Dreaming is still a little understood phenomenon of the human psyche, although there is a steady but slow uptake in releated research [1,2]. Especially its purpose remains somewhat a mystery and many suggetions have been put forth [3,4].In this short article I do not want to sketch a general answer...
Read moreTalking counterfactual #5: BlenderBot 3
November 28, 2022
It has been quite some time, but here we are again back at the counterfactual game and today I have a premiere for you. We have a guest who will be the second time on the show presenting an improved version of itself. I’m talking of no other than BlenderBot...
Read moreTalking counterfactual #4: BIG SCIENCE’s T0pp
Dezember 23, 2021
Throughout the year 2021 many new interesting models were released (for instance GPT-J [1] or Megatron-Turing NLG [2]) others got updated (ERNIE 3.0 [3], PLATO-XL [4] and blenderbot 2.0 [5]), so it is quite difficult to pick the next target for our talking counterfactual series. Though we have been looking...
Read moreTalking counterfactual #3: OpenAI’s GPT-3
Oktober 27, 2021
I assume GPT-3 from OpenAI does not need an introduction. Released in summer 2020, it is a 175 billion parameter transformer based behemoth delivering state of the art performance in many NLP tasks. So, it’s an obvious choice to try out and see how it handles counterfactual conditionals. Other than...
Read moreTalking counterfactual #2: Microsoft’s DialoGPT
Juni 9, 2021
In episode two of our new series, we take on Microsoft’s DialoGPT. It is a GPT-2 transformer-style architecture trained on 147 million Reddit discussion threads released in November 2019 [1,2]. Now, Reddit might not be the best spot to train your AI (so we heard 😉), but it is one...
Read moreTalking counterfactual #1: facebook’s BlenderBot
Mai 28, 2021
With crass.ai getting ready for public beta, we thought it’s time to take it for a spin and have a dance with some of the recent LLMs (large language models).As a starting point, we chose facebook’s BlenderBot (for convenience followingly written without the second capital “B”). The Blenderbot is a...
Read more