alexcz

Yeah but I think this accelerator is more meant for a small scale deployment. This could probably serve a medium moe llm model at reasonable tk/s plus you could in addition to the llm also run a couple of other models on it in parallel like an embedding, reranker, STT, TTS, decision, image, video, … if you deploy this for a small business they might make use of all of these models but not all at the same time and not with 100s of users but this card could bring a serious broad range of AI capabilities for them.

could not agree more one pet peeve of mine is that using strict json output also does not work on all the endpoints of the models so I have to go through them one by one try it out and then only whitelist them.

It gets even more interesting if you take into account how the satellites know where they are. Around the world there are fundamental stations: https://en.wikipedia.org/wiki/Fundamental_station

I had the opportunity to visit one. Basically they measure their own position in relation to each other. They do that with Very-long-baseline interferometry, basically what is the time difference of quasar radio signals hitting their Radio telescopes. The things they account for is wild like local gravity field a couple of super prices atomic clocks etc. they then laser range find Satellites (all not only gps) which is a „fun“ summer student job at least at the one that I visited.

Really like it, only thing some of the cells can be jittery and rapidly switch back and forth between two symbols, making for an unpleasant effect, maybe there is a way to smooth this?