Latest

Researchers Discover Emergent Linear Structures in How LLMs Represent Truth

LLMs contain a specific "truth direction" denoting factual truth values.

Tsinghua University: Inverting Transformers Significantly Improves Time Series Forecasting

Inverting Transformer architecture for time series forecasting improves performance

ALMT: Using text to narrow focus in multimodal sentiment analysis improves performance

A new way to filter multimodal data for sentiment analysis using text guidance improves performance

Thought Propagation: Teaching LLMs to Solve Complex Reasoning Tasks with Analogies

Humans solve complex and novel problems using analogies. Can LLMs do the same?

There are many ways to design roundabouts. Can AI choose the best one?

Probabilistic generative modeling for procedural roundabout generation for developing countries

Researchers: Low-Resource Languages Can Easily Jailbreak LLMs

These results clearly demonstrate safety mechanisms do not properly generalize across languages

Decoding Speech from Brain Waves - A Breakthrough in Brain-Computer Interfaces

Researchers from Meta have discovered how to turn brain waves into speech using noninvasive methods like EEG and MEG