Record scratch—Google’s Lyria 3 AI music model is coming to Gemini today

Google DeepMind has been tinkering with Lyria for a while now, offering limited access in developer-oriented products like ...
IEEE

Fully Exploiting Every Real Sample: SuperPixel Sample Gradient Model Stealing

Abstract: Model stealing (MS) involves querying and observing the output of a machine learning model to steal its capabilities. The quality of queried data is crucial, yet obtaining a large amount of ...
IEEE

Estimating Individual Building Heights by Integrating Spaceborne LiDAR and Multisource Remote Sensing Data: A CNN–Transformer Model and a Semi-Supervised Sample Augmentation ...

Abstract: Building height is a key metric in transitioning urban analysis from 2-D to 3-D perspectives. It serves as a fundamental indicator for assessing urban development, population density, and ...
GitHub

A Generative Foundation Reward Model (GRAM)

This repository contains the code and released models for our paper GRAM: A Generative Foundation Reward Model for Reward Generalization 📝. We propose a more effective approach to reward model ...
GitHub

Revolutionizing Reinforcement Learning Framework for Diffusion Large Language Models

We propose TraceRL, a trajectory-aware reinforcement learning method for diffusion language models, which demonstrates the best performance among RL approaches for DLMs. We also introduce a ...