Toggle light / dark theme

Get the latest international news and world events from around the world.

Log in for authorized contributors

Reinforcement learning boosts reasoning skills in new diffusion-based language model d1

A team of AI researchers at the University of California, Los Angeles, working with a colleague from Meta AI, has introduced d1, a diffusion-large-language-model-based framework that has been improved through the use of reinforcement learning. The group posted a paper describing their work and features of the new framework on the arXiv preprint server.

Over the past couple of years, the use of LLMs has skyrocketed, with millions of people the world over using AI apps for a wide variety of applications. This has led to an associated need for large amounts of electricity to power data centers running the computer-intensive applications. Researchers have been looking for other ways to provide AI services to the user community. One such approach involves the use of dLLMs as either a replacement or complementary approach.

Diffusion-based LLMs (dLLMs) are AI models that arrive at answers differently than LLMs. Instead of taking the autoregressive approach, they use diffusion to find answers. Such models were originally used to generate images—they were taught how to do so by adding overwhelming noise to an image and then training the model to reverse the process until nothing was left but the original image.

Clustered neurons in bat midbrain encode categories of vocalizations, study finds

The ability to quickly recognize sounds, particularly the vocalizations made by other animals, is known to contribute to the survival of a wide range of species. This ability is supported by a process known as categorical perception, which entails the transformation of continuous auditory input (e.g., gradual changes in pitch or tone) into distinct categories (i.e., vocalizations that mean something specific).

Various past studies have tried to shed light on the neural underpinnings of categorical perception and the categorization of vocalizations. While they broadly identified some that could play a part in these abilities, the precise processes through which animals categorize their peer’s categorizations have not yet been fully elucidated.

Researchers at Johns Hopkins University recently carried out a study investigating how vocalizations are represented in the brain of big brown bats, which are scientifically known as Eptesicus fuscus. Their findings, published in Nature Neuroscience, suggest that the categories of vocalizations are encoded in the bat midbrain.

Gravity is Result of Computational Process within Our Universe, Physicist Says

“My findings in this study fit with the thought that the Universe might work like a giant computer, or our reality is a simulated construct,” Dr. Vopson said.

“Just like computers try to save space and run more efficiently, the Universe might be doing the same.”

“It’s a new way to think about gravity — not just as a pull, but as something that happens when the Universe is trying to stay organized.”