They are also useful for maintaining orbits (like the ISS), and could potentially be useful in long duration missions such as manned interplanetary transfers. If it takes months to get to Mars anyway, low thrust won't matter if it means you get more total delta-v, and able to reduce the travel duration anyway

Blame politicians for not stopping number spoofing. It's insane that someone can pay to use existing numbers, and today it's only really used by scammers. If all these calls showed up as coming from a random foreign country, no one would ever answer them. There's a very easy solution to this.

I used it for boilerplate work. Like if I need to write a model in typescript with a ton of class-validator annotations, it's pretty good at filling those in

Some employees though, not so much

Ding ding ding.

I didn't. I must admit, I don't really know how it all works and am mostly just blindly picking my way through, appreciating what works at this moment in time.

Think of it like torrents. You need tj search a site to get the magnet link (indexer) Then you need to connect to the dht network (provider). Then you need a download client (sabnzbd+). It’s not exactly the same, because both the provider and the indexed usually charge money to access, but in some places it’s not illegal. The illegal part is sharing in some jurisdictions

Have you tried one of the google search alternatives? kagi Quant Ecosia duckduckgo Of course they can not compensate for the shithole the whole internet has become

No, it just makes it boring straight

This, flat GDP really only accurately measures US economy and maybe the Euro zone. With others it becomes increasingly useless in really estimating anything concrete about that economy. Doubly so if they are sanctioned.

Another tool for bulk downloading, with 811 stars for the record: https://github.com/treetrum/amazon-kindle-bulk-downloader

Deepseek continue its open source releases DeepEP is a communication library tailored for Mixture-of-Experts (MoE) and expert parallelism (EP). It provides high-throughput and low-latency all-to-all GPU kernels, which are also as known as MoE dispatch and combine. The library also supports low-precision operations, including FP8. To align with the group-limited gating algorithm proposed in the DeepSeek-V3 paper, DeepEP offers a set of kernels optimized for asymmetric-domain bandwidth forwarding, such as forwarding data from NVLink domain to RDMA domain. These kernels deliver high throughput, making them suitable for both training and inference prefilling tasks. Additionally, they support SM (Streaming Multiprocessors) number control. For latency-sensitive inference decoding, DeepEP includes a set of low-latency kernels with pure RDMA to minimize delays. The library also introduces a hook-based communication-computation overlapping method that does not occupy any SM resource. Notice: the implementation in this library may have some slight differences from the DeepSeek-V3 paper.