In our recent paper, we explore how populations of deep reinforcement learning (deep RL) agents can learn microeconomic behaviours, such as production, consumption, and trading of goods. We find that artificial agents learn to make economically rational decisions about production, consumption, and prices, and react appropriately to supply and demand changes. The population converges to…
Photo by Mathew Schwartz on UnsplashIn the previous article, we discussed the surprising behavior of data in higher dimensions. We found that volume tends to accumulate in the corners of spaces in a strange way, and we simulated a hypersphere inscribed inside a hypercube to investigate this, observing an interesting decrease in their volume ratio…
Large language models have shown notable achievements in executing instructions, multi-turn conversations, and image-based question-answering tasks. These models include Flamingo, GPT-4V, and Gemini. The fast development of open-source Large Language Models, such as LLaMA and Vicuna, has greatly accelerated the evolution of open-source vision language models. These advancements mainly center on improving visual understanding by…
Understand batch processing from business and technical perspective Photo by Dannie Sorum on UnsplashWe live in a world where every human interaction becomes an event in the system, whether it’s purchasing clothes online or in-store, scrolling social media, or taking an Uber. Unsurprisingly, all these events are processed in one way or the other.…
In the rapidly evolving digital imagery and 3D representation landscape, a new milestone is set by the innovative fusion of 3D Generative Adversarial Networks (GANs) with diffusion models. The significance of this development lies in its ability to address longstanding challenges in the field, particularly the scarcity of 3D training data and the complexities associated…
Reinforcement learning (RL) has made tremendous progress in recent years towards addressing real-life problems – and offline RL made it even more practical. Instead of direct interactions with the environment, we can now train many algorithms from a single pre-recorded dataset. However, we lose the practical advantages in data-efficiency of offline RL when we evaluate…
Exploring Typeform Alternatives In the market of digital surveys and forms, Typeform set a high standard with its intuitive design and user experience. The European startup quickly became a go-to for businesses and individuals looking to gather information efficiently. However, the requirements one may have from a data collection exercise may be diverse, so there…
Gain intuition behind acceleration training techniques in neural networks D eep learning made a gigantic step in the world of artificial intelligence. At the current moment, neural networks outperform other types of algorithms on non-tabular data: images, videos, audio, etc. Deep learning models usually have a strong complexity and come up with millions or even…
Image by Author
What a year, ey?
At the start of 2023, people were struggling to keep up with the tech world. One day this was getting released, the next day a competitor came out with something else, and then you heard about something new. It was a lot.
But the momentum continued…
In virtual reality and 3D modeling, constructing dynamic, high-fidelity digital human representations from limited data sources, such as single-view videos, presents a significant challenge. This task demands an intricate balance between achieving detailed and accurate digital representations and the computational efficiency required for real-time applications. Traditional methods often grapple with rendering speeds and model fidelity…