文章预览
LG - 机器学习 CV - 计算机视觉 CL - 计算与语言 AI - 人工智能 1、[LG] Diffusion Models Are Real-Time Game Engines 2、[LG] The Benefits of Balance：From Information Projections to Variance Reduction 3、[CL] Relationships are Complicated! An Analysis of Relationships Between Datasets on the Web 4、[LG] Generative Verifiers：Reward Modeling as Next-Token Prediction 5、[LG] The Mamba in the Llama：Distilling and Accelerating Hybrid Models 摘要：扩散模型用作实时游戏引擎、平衡化的好处:从信息预测到方差减低、网络数据集间关系分析、生成式验证器、将大型Transformer语言模型迁移蒸馏到线性RNN模型 1、[LG] Diffusion Models Are Real-Time Game Engines D Valevski, Y Leviathan, M Arar, S Fruchter [Google Research] 扩散模型用作实时游戏引擎 要点: 异步扩散模型可以以高质量和实时交互来模拟复杂的交互环境，如视频游戏。 他们提出GameNGen，一个可以以20帧每秒的速
………………………………
