Tootfinder

1. Plan going to Opole, via Kościan.
2. When you enter the train to Kościan, you discover that the change to Opole is delayed 15 minutes already. Consider changing in Leszno instead; if the delay increases, you'd have more options there.
3. Discover that there aren't any more options in Leszno today. Your change is delayed 30 minutes already. Return the reservations, and take one the other way, to Poznań instead.
4. Train station in Kościan. The displays aren't showing any delays, trains are announced normally. Tell people about the delays, so they won't stand in the -10°C waiting for the train to arrive.
5. Take the train to Poznań, and try to figure out what to do next.
6. Discover that the only reasonable choice going forward is Inowrocław: no delays and good return connection. It's the same train, so take another reservation. Your current seat is already taken there, so move elsewhere.
7. Your train should be followed by another one in the same direction, that departs from Poznań 6 minutes later. However, your train ends up waiting for another delayed train, so the other train goes first. The delay further increases as your train needs to slow down after the other train.
8. Reach Inowrocław 10 minutes later. That's not a problem, since you didn't have enough to see for all the time there anyway.
9. Discover that the town is more interesting than you thought, and you'd use more time.
10. When you almost get to the station, discover that your train is 10 minutes late. Not that you have any use for that time at this point.
11. When you're at the station, the train keeps increasing delay while waiting at the previous station, in Bydgoszcz. The station displays are completely useless, as they show only a random subset of regional trains, for no apparent reason. The announcements include all trains, but are rarely given.
12. The delay keeps increasing. Start thinking about getting a reservation for the next train to Poznań, in case it arrived first. You can't return the reservation after the planned departure time, and you can't have two reservations simultaneously, so reserve the seat from Mogilno, the next station.
13. The next train arrives first. While on board, you discover that you're not going to have any train home for 1.5 hr. Take another seat reservation to Leszno, where you can change into a suburban train and get home 15 minutes earlier than from Poznań. This time, your seat is still free.
14. The train departs 15 minutes delayed from Poznań. After all, you're changing trains in Kościan.
So I was going to go south, to Opole, via Kościan. Instead, I've ended up slingshotting north to Inowrocław, and getting back home via the same train as if I were in Opole.
#rail

UrbanFM: Scaling Urban Spatio-Temporal Foundation Models
Wei Chen, Yuqian Wu, Junle Chen, Xiaofang Zhou, Yuxuan Liang
https://arxiv.org/abs/2602.20677 https://arxiv.org/pdf/2602.20677 https://arxiv.org/html/2602.20677
arXiv:2602.20677v1 Announce Type: new
Abstract: Urban systems, as dynamic complex systems, continuously generate spatio-temporal data streams that encode the fundamental laws of human mobility and city evolution. While AI for Science has witnessed the transformative power of foundation models in disciplines like genomics and meteorology, urban computing remains fragmented due to "scenario-specific" models, which are overfitted to specific regions or tasks, hindering their generalizability. To bridge this gap and advance spatio-temporal foundation models for urban systems, we adopt scaling as the central perspective and systematically investigate two key questions: what to scale and how to scale. Grounded in first-principles analysis, we identify three critical dimensions: heterogeneity, correlation, and dynamics, aligning these principles with the fundamental scientific properties of urban spatio-temporal data. Specifically, to address heterogeneity through data scaling, we construct WorldST. This billion-scale corpus standardizes diverse physical signals, such as traffic flow and speed, from over 100 global cities into a unified data format. To enable computation scaling for modeling correlations, we introduce the MiniST unit, a novel split mechanism that discretizes continuous spatio-temporal fields into learnable computational units to unify representations of grid-based and sensor-based observations. Finally, addressing dynamics via architecture scaling, we propose UrbanFM, a minimalist self-attention architecture designed with limited inductive biases to autonomously learn dynamic spatio-temporal dependencies from massive data. Furthermore, we establish EvalST, the largest-scale urban spatio-temporal benchmark to date. Extensive experiments demonstrate that UrbanFM achieves remarkable zero-shot generalization across unseen cities and tasks, marking a pivotal first step toward large-scale urban spatio-temporal foundation models.
toXiv_bot_toot

Google is testing a new ad format in AI Mode that displays retailers selling the products a user is searching for in a Sponsored box (Vidhya Srinivasan/Google Ads & Commerce Blog)
https://blog.google/products/ads-commerce/agentic-commerce…

New tech and tools for retailers to succeed in an agentic shopping era
An overview of Google’s new open standard for agentic commerce and AI tools to help retailers connect with shoppers and drive sales.

Market Value 💵
市场价值 💵
📷 Nikon F4E
🎞️ ERA 100, expired 1993
#filmphotography #Photography #blackandwhite

For fans of #Steam games where you can build -really- broken mechanics like Balatro or Slay the Spire (#deckbuilders, #roguelikes, etc. - "Make the numbers go UP!"), LOOTPLOT has been *extremely* satisfying for me
It's like: "what if Incredible Machine, but Ballionaire" - you randomly get access to a bunch of items that each have their own mechanics and trigger/play off each other, which you build out on a grid - but the grid itself can be manipulated by the items. Some of the interactions are absolutely wild.
Plays a little rough on CrossOver - I think the dynamic scaling plays havoc with texture calculation - but is still enjoyable all the same. (It's written in LÖVE - I really hope the author releases a macOS build!)
The sale price of $3.49 is crazy good for the enjoyment value I've gotten out of it
https://store.steampowered.com/app/3057190/LOOTPLOT/