Mastodawn

https://winbuzzer.com/2026/06/11/google-diffusiongemma-trades-quality-for-local-ai-speed-xcxwbn/

Google has introduced DiffusionGemma to speed local AI output through parallel text diffusion, but lower quality than Gemma 4 keeps trade-offs visible.

#AI #DiffusionGemma #TextDiffusion #Google #GoogleAI #AIModels #OpenSourceAI #OnDeviceAI #AIResearch

AI Sparkup Nov 10, 2025

표준 LLM을 넘어서: 5가지 대안 아키텍처 둘러보기

GPT와 Claude를 넘어서는 5가지 대안 LLM 아키텍처 소개. Linear Attention으로 메모리 75% 절감, Diffusion으로 병렬 생성, 코드 실행을 시뮬레이션하는 Code World Model까지.

https://aisparkup.com/posts/6342

N-gated Hacker News Oct 20, 2025

🎉 Wow, who knew #BERT was just a one-step wonder in the mystical art of text diffusion? 😮✨ Apparently, all it takes is a sprinkle of #AI jargon and voilà, we've reinvented the wheel of #transformers. 🚀🔥
https://nathan.rs/posts/roberta-diffusion/ #textdiffusion #innovation #technology #HackerNews #ngated

BERT is just a Single Text Diffusion Step

A while back, Google DeepMind unveiled Gemini Diffusion, an experimental language model that generates text using diffusion. Unlike traditional GPT-style models that generate one word at a time, Gemini Diffusion creates whole blocks of text by refining random noise step-by-step. I read the paper Large Language Diffusion Models and was surprised to find that discrete language diffusion is just a generalization of masked language modeling (MLM), something we’ve been doing since 2018. The first thought I had was, “can we finetune a BERT-like model to do text generation?” I decided to try a quick proof of concept out of curiosity.

nathan.rs

Hacker News Oct 20, 2025

BERT Is Just a Single Text Diffusion Step

https://nathan.rs/posts/roberta-diffusion/

#HackerNews #BERT #TextDiffusion #AI #NLP #MachineLearning

BERT is just a Single Text Diffusion Step

A while back, Google DeepMind unveiled Gemini Diffusion, an experimental language model that generates text using diffusion. Unlike traditional GPT-style models that generate one word at a time, Gemini Diffusion creates whole blocks of text by refining random noise step-by-step. I read the paper Large Language Diffusion Models and was surprised to find that discrete language diffusion is just a generalization of masked language modeling (MLM), something we’ve been doing since 2018. The first thought I had was, “can we finetune a BERT-like model to do text generation?” I decided to try a quick proof of concept out of curiosity.

nathan.rs