XDA Developers on MSN

6 local LLMs I've used that prove they're not just smaller versions of cloud models

These local LLMs are changing the game in lots of fun ways.
GitHub

ARDY: Autoregressive Diffusion with Hybrid Representation for Interactive Human Motion Generation

ARDY is an autoregressive diffusion model designed for interactive motion generation, supporting online text prompting and flexible long-horizon kinematic constraints (root paths/waypoints, full-body ...
Geeky Gadgets

Google’s New Diffusion Gemma Changes How AI Processes Language

Google’s Diffusion Gemma introduces a bold shift in AI language modeling by adopting a diffusion-based architecture that processes tokens in parallel, rather than sequentially. As explained by Prompt ...
marktechpost

Google DeepMind Releases Gemma 4 12B: An Encoder-Free Multimodal Model with Native audio that runs on a 16 GB laptop

Gemma 4 12B is a 12-billion-parameter decoder-only transformer. It handles text, images, audio, and video natively. There are no separate vision or audio encoders. The decoder uses the same structure ...
IEEE

StigDiff: Stigmergy Diffusion for Signal Compensation in Multi-Agent Localization

Abstract: Multi-agent localization faces the problem of signal loss or attenuation. To address the signal compensation problem, we propose a novel diffusion model for multi-agent signal compensation, ...
GitHub

steering_one-step_diffusion_model_with_fidelity-rich_decoder_for_fast_image_comp.md

description [AAAI 2026][Image Generation][Image Compression] This paper proposes SODEC, a one-step diffusion-based image compression model that injects the prior of a high-fidelity VAE decoder into ...
The New York Times

White House Considers Vetting A.I. Models Before They Are Released

The Trump administration, which took a noninterventionist approach to artificial intelligence, is now discussing imposing oversight on A.I. models before they are made publicly available. By Tripp ...
Hacker

Building a Transformer From Scratch in Annotated PyTorch

A complete walkthrough of implementing the original Attention Is All You Need encoder-decoder Transformer—no torch. nn.Transformer, no shortcuts. The 2017 paper "Attention Is All You Need" by Vaswani ...
Geeky Gadgets

Mercury 2 : World’s Fastest Reasoning AI Model Built for Production Applications

Mercury 2, the first diffusion-based reasoning large language model, introduces a new approach to token generation by refining multiple tokens in parallel rather than sequentially. This shift enables ...
EurekAlert!

Latent diffusion model provides efficient and high-quality paraphrase

LDP consists of a diffusion modeling for encoded text space of an off-the-shelf pre-trained encoder and decoder, the diffusion process can be intervened by additional controller . Paraphrase ...
Hacker

Diffusion Models Are Real-Time Game Engines

We present GameNGen, the first game engine powered entirely by a neural model that also enables real-time interaction with a complex environment over long trajectories at high quality. When trained on ...