More

    Science

    The Unfaithful Nature of Chain-of-Thought Explanations in Large Language Models

    A Study Reveals How Misleading Explanations Can Increase Trust in AI Systems Without Ensuring Their Safety Chain-of-thought (CoT) explanations produced by large language models (LLMs)...

    Vcc: A Breakthrough in Scaling Transformers to Handle Ultra-Long Sequences

    Prioritizing Important Tokens to Achieve Over 3x Efficiency Improvement for 4K to 128K Token Lengths Vcc (VIP-token centric compression) tackles the challenge of efficiently processing...

    Evaluating the Accuracy of AI-Generated Code with EvalPlus

    A Rigorous Evaluation Framework for Code Synthesis with Large Language Models Existing code evaluation datasets may not fully assess the functional correctness of code generated...

    NVIDIA Unveils Cutting-Edge AI Research at SIGGRAPH 2023

    Groundbreaking advancements in generative AI, neural graphics, and realistic simulations NVIDIA Research presents around 20 papers on generative AI and neural graphics, in collaboration with...

    LLaMA-Adapter V2: Next-Gen Parameter-Efficient Visual Instruction Model

    Enhanced multi-modal reasoning and visual instruction-following capabilities with minimal additional parameters LLaMA-Adapter V2 unlocks more learnable parameters and introduces an early fusion strategy for better...

    Unlimiformer: A Breakthrough in Long-Range Transformers with Unlimited Length Input

    New approach offloads attention computation to a single k-nearest-neighbor index, enabling extremely long input sequences Unlimiformer can wrap any existing pretrained encoder-decoder transformer, allowing it...

    Assessing the Labor Market Impact of Large Language Models

    Study reveals 80% of the U.S. workforce may see at least 10% of their tasks affected by LLMs, and higher-wage jobs face greater exposure Around...

    DeepMind Trains Miniature Humanoid Robots to Play Soccer

    Google's DeepMind uses Deep Reinforcement Learning to teach robots soccer skills and strategies DeepMind trained miniature humanoid robots with 20 controllable joints to play soccer...

    The Eye’s Incredible Data Processing: How Little Information Reaches the Brain

    A closer look at the retina's role in compressing and transmitting visual data to the brain The retina compresses a significant amount of visual information...

    Enhancing Language Models with Self-Notes for Improved Reasoning and Memorization

    A novel approach extends memory and enables multi-step reasoning in large language models Self-Notes method addresses limitations in context memory and multi-step reasoning in large...

    AI Pioneer Geoffrey Hinton Departs Google, Expresses Concerns about Technology

    Deep learning visionary leaves tech giant after a decade, voicing new concerns over AI's potential risks Geoffrey Hinton, a pioneer of deep learning and AI,...

    Exploring Self-Supervised Vision Transformers: A Comparative Study on CL and MIM

    Analyzing the Properties of Contrastive Learning and Masked Image Modeling in Vision Transformers and Their Potential Synergy The study compares self-supervised learning methods Contrastive Learning...