RLHF: Teaching robots right and wrong

On the cover: Screenshot from the movie “Ron’s gone wrong” where a kid teaches his pet robot how to behave socially The term “RLHF” or Reinforcement Learning from Human Feedback has become popular due to Conversational AI models like ChatGPT. How...

ChatGPT - The Conversational Wizard

On the cover: ChatGPT’s response to the prompt “What is the meaning of life?” As the AI war between Google and Microsoft is brewing, it’s high time we understand what’s going on under these conversational AI models. If you have tried ChatGPT, you...

Guided Diffusion Models - Part 2

On the cover: Midjourney’s creation for the prompt “Sun Goddess artful” We saw how we could train a generative diffusion model in the previous post. But what fun is it if you can’t generate something of your choice. Welcome to guidance for diffus...

Guided Diffusion Models - Part 1

On the cover: Midjourney’s creation for the prompt “Astronauts exploring an alien red planet” Diffusion models are all the rage these days. Why do we need these new models for image generation? In fact, we have GANs, VAEs and Flow models that hav...