According to AOMedia's announcement, AV2 is set for release at the end of 2025. The group highlights several improvements over AV1, including significantly better compression performance (with no ...

"Understanding Large Models for Humanities Students (1.0)" is written by Penny Liang, offering a simplified perspective to ...

Using light as a decoder for a diffusion-based image generator, scientists show that improvements can be made to make AI ...

Scientists at NYU Abu Dhabi (NYUAD) have developed an artificial intelligence (AI) model that can forecast solar wind speeds ...

In the rapid evolution of multimodal large models, the visual module has always been a key cornerstone supporting the entire ...

When researchers are building large language models (LLMs), they aim to maximize performance under a particular computational ...

A new AI model from NYU Abu Dhabi predicts solar wind days in advance with far greater accuracy than existing methods. By analyzing ultraviolet solar images, it could help protect satellites, ...

MAE, a self-supervised learning framework designed to automate 3D segmentation of plant organs from point cloud data.

Recent advances in high-throughput microbiome profiling have generated expansive data sets that offer unprecedented ...

文｜博阳编辑｜可君今年6月，阶跃星辰首席科学家张祥雨在访谈中谈及了他在近两年模型训练中遇到的最大困境——多模态AI的内部，一直有一场“内战”。具体是，在大一统多模态模型训练中，视觉的“理解”与“生成”能力可以共存，却很少协作，甚至时常内耗 ...

张祥雨的思维链方案确实有其道理，但它主要解决的是单次推理的复杂度问题。而北大团队却发现了一个更根本的问题：理解和生成的训练目标本身就是割裂的。即使引入思维链，两个模块依然是在追求不同的KPI。