Tootfinder

Global Energy Monitor: over 97 GW of gas-fired power in the US pipeline were explicitly earmarked for data centers in 2025, compared with just 4 GW in 2024 (Molly Taft/Wired)
https://www.wired.com/story/data-centers-are-driving-a-us-gas-boom/

Data Centers Are Driving a US Gas Boom
Gas projects in the US pipeline explicitly linked to data centers increased by almost 25 times over the past two years, according to new research from Global Energy Monitor.

Scaling Vision Transformers: Evaluating DeepSpeed for Image-Centric Workloads
Huy Trinh, Rebecca Ma, Zeqi Yu, Tahsin Reza
https://arxiv.org/abs/2602.21081 https://arxiv.org/pdf/2602.21081 https://arxiv.org/html/2602.21081
arXiv:2602.21081v1 Announce Type: new
Abstract: Vision Transformers (ViTs) have demonstrated remarkable potential in image processing tasks by utilizing self-attention mechanisms to capture global relationships within data. However, their scalability is hindered by significant computational and memory demands, especially for large-scale models with many parameters. This study aims to leverage DeepSpeed, a highly efficient distributed training framework that is commonly used for language models, to enhance the scalability and performance of ViTs. We evaluate intra- and inter-node training efficiency across multiple GPU configurations on various datasets like CIFAR-10 and CIFAR-100, exploring the impact of distributed data parallelism on training speed, communication overhead, and overall scalability (strong and weak scaling). By systematically varying software parameters, such as batch size and gradient accumulation, we identify key factors influencing performance of distributed training. The experiments in this study provide a foundational basis for applying DeepSpeed to image-related tasks. Future work will extend these investigations to deepen our understanding of DeepSpeed's limitations and explore strategies for optimizing distributed training pipelines for Vision Transformers.
toXiv_bot_toot

There are a lot of takeaways from this:
1. Organizing locally gives you a massive advantage because you will always know your local area better than ICE ever can.
2. Be agile. You can always change tactics faster than a centralized organization.
3. Organize now. The sooner you build your networks, the sooner you can learn.
4. Identify ICE facilities and organize monitoring them directly.
But I think the most interesting one that's not explicitly in there, one that's hinted at the last one, is to go on the offensive. ICE is already afraid. If we all take the anger we have at the murder of #ReneGood, find the local ICE facility that they'll stage from, and bring that anger to #OccupyICE we might be able to just shut the whole thing down preemptively. Completely stop all ICE operations across the US. If they want to fight, they can fight *with everyone, all at once.*
Shut down their ability to operate at all. They have a logistics pipeline. They need cars, they need oil in those cars, they need to be able to move those cars to target areas. They also need money to pay those agents. All of those can be disrupted.
The regime needs your money and labor to maintain the illusion of legitimacy. They chose a bad time because you can hit both of those at once *right now* with a combination of #GeneralStrike and #TaxStrike, and then #BoycottEverything.
The regime is weaker than it's ever been. It's flailing. Their own base is demanding the release of the #EpsteinFiles. Their last gasp attempt to prevent the radical change that's coming is just to ethnically cleanse the US back to the 50's (which is what they always meant by "Make America Great Again"). Trump will do anything to stay in power, even if it means killing everyone on Earth in the process. But Americans can end it now by going on the offensive.
Now is the time.
#USPol