Tootfinder

The Most AI-Exposed Counties in America Are Not Where You Think
https://jakeprokopets.substack.com/p/why-the-most-ai-exposed-counties
Here is a county-level AI displacement model across all 3,204 US counties.
The Top 5 most exposed counties are a…

Why the Most AI-Exposed Counties in America Are Not Where You Think
A county-level model of AI workforce displacement built on peer-reviewed methodology, BLS data, and real deployment evidence. The findings surprised me.

Kraken launches what it claims are the first regulated perpetual futures contracts based on tokenized stocks, available to non-US users and trading 24/7 (Krisztian Sandor/CoinDesk)
https://www.coindesk.com/business/2026/02/

Kraken brings crypto-style, 24/7 perpetuals trading for tokenized U.S. stocks
The crypto exchange is bringing U.S. stocks closer to the crypto world, letting users trade derivatives of tokenized stocks around-the-clock and with 20x leverage.

Contextualizing the Proposed SECURE Data Act in the State Privacy Landscape
https://fpf.org/blog/contextualizing-the-proposed-secure-data-act-in-the-state-privacy-landscape/
@…

sp_primary_school: Primary school dynamic contacts (2009)
Two temporal networks of contacts among students and teachers at a primary school in Lyon, France, on consecutive days of in October 2009. Each network accumulates all contacts over the course of a single day; contacts were sampled at 20-second intervals.
This network has 242 nodes and 125773 edges.
Tags: Social, Offline, Weighted, Temporal, Metadata

Liquid ass: now with 100% more context menu.
#apple

Untied Ulysses: Memory-Efficient Context Parallelism via Headwise Chunking
Ravi Ghadia, Maksim Abraham, Sergei Vorobyov, Max Ryabinin
https://arxiv.org/abs/2602.21196 https://arxiv.org/pdf/2602.21196 https://arxiv.org/html/2602.21196
arXiv:2602.21196v1 Announce Type: new
Abstract: Efficiently processing long sequences with Transformer models usually requires splitting the computations across accelerators via context parallelism. The dominant approaches in this family of methods, such as Ring Attention or DeepSpeed Ulysses, enable scaling over the context dimension but do not focus on memory efficiency, which limits the sequence lengths they can support. More advanced techniques, such as Fully Pipelined Distributed Transformer or activation offloading, can further extend the possible context length at the cost of training throughput. In this paper, we present UPipe, a simple yet effective context parallelism technique that performs fine-grained chunking at the attention head level. This technique significantly reduces the activation memory usage of self-attention, breaking the activation memory barrier and unlocking much longer context lengths. Our approach reduces intermediate tensor memory usage in the attention layer by as much as 87.5$\%$ for 32B Transformers, while matching previous context parallelism techniques in terms of training speed. UPipe can support the context length of 5M tokens when training Llama3-8B on a single 8$\times$H100 node, improving upon prior methods by over 25$\%$.
toXiv_bot_toot

Sources: Meta aims to enter the stablecoin space in H2, with plans to integrate a vendor to administer stablecoin-backed payments and implement a new wallet (Ian Allison/CoinDesk)
https://www.coindesk.com/business/2026/02/

Mark Zuckerberg's Meta is planning stablecoin comeback in the second half amid U.S. regulatory shift
The Facebook owner’s stablecoin integration involves planning a third party vendor to help administer stablecoin-based payments and a new wallet to be implemented, sources said.

Circle shares fell as much as 18%, and Coinbase dropped about 8%, after a draft of the US Clarity Act raised the prospect of strict limits on stablecoin yield (CoinDesk)
https://www.coindesk.com/markets/2026/03/2

CRCL, COIN news: Circle, Coinbase tumbles as regulators move to ban interest on stablecoins
The latest version of the Clarity Act is pressuring stocks as it would restrict stablecoin rewards.

sp_office: Workplace contacts (2013)
A temporal network of contacts between individuals, measured in an office building in France, from June 24 to July 3, 2013.
This network has 92 nodes and 9827 edges.
Tags: Social, Offline, Weighted, Temporal, Metadata
https://networks.skewed.de/net/sp_offi