iiinova — where your digital self keeps evolving

Cullen Miller @cullenmiller.bsky.social · 30/07/2026

This is incredible. How can I learn more about this? Are you planning to publish or release anything on this beyond the streams? Dying to learn more

010

Cullen Miller @cullenmiller.bsky.social · 23/03/2026

Louis Mosley (grandson of Oswald Mosley - yes,THAT Oswald Mosley) is the head of palantir uk. He’s central to the company’s expansion within the UK government, overseeing major contracts with the NHS and the Ministry of Defence

063

Cullen Miller @cullenmiller.bsky.social · 05/10/2025

Been Jünger-curious for a while. Which book is the best to start with?

040

Cullen Miller @cullenmiller.bsky.social · 05/04/2025

both discuss the irony of the US creating ARPANet through central planning and the Soviets creating OGAS via competing agencies; the complete inverse of the values they were projecting outward.

020

Cullen Miller @cullenmiller.bsky.social · 05/04/2025

"how not to network a nation" and "internyet" web.mit.edu/slava/homepa... mitpress.mit.edu/978026253466...

web.mit.edu

110

Cullen Miller @cullenmiller.bsky.social · 31/10/2024

The PD12M paper details our methodology and presents a framework for others building public datasets. We welcome feedback from the broader community as we work to expand access to high-quality, responsibly-sourced training data. 📝 Preprint paper : arxiv.org/abs/2410.23144

arxiv.org

Public Domain 12M: A Highly Aesthetic Image-Text Dataset with Novel Governance Mechanisms

We present Public Domain 12M (PD12M), a dataset of 12.4 million high-quality public domain and CC0-licensed images with synthetic captions, designed for training text-to-image models. PD12M is the lar...

010

Cullen Miller @cullenmiller.bsky.social · 31/10/2024

Today Spawning is releasing PD12M, a dataset of 12.4M public domain images and synthetic captions, alongside research detailing a novel approach to dataset governance. Dataset 🔗: source.plus/pd12m

source.plus

PD12M | Source.Plus

Search, curate, and enrich media collections for AI training using the Source.Plus marketplace. Safe, consenting, high-quality training data. Public domain datasets for model fine-tuning.

140

Cullen Miller @cullenmiller.bsky.social · 15/10/2023

Went to a vintage computer exhibition and am proud to say that I can still hear the 15.75kHz emission from CRT monitors

050

Cullen Miller @cullenmiller.bsky.social · 11/10/2023

using light mode = guaranteed serial killer

000

Cullen Miller @cullenmiller.bsky.social · 13/06/2023

Big guy is back in the dopamine trough.

090