The DeepSeek Code
For years it was taken for granted that competing in artificial intelligence required the budget of a nation state. A small team in a provincial city proved otherwise. The story does not begin in a famous lab or a famous university. It begins in a quantitative investment fund: people used to hunting for patterns in price series, fighting over milliseconds, and working with the resources actually available rather than the ones that would have been convenient. Out of that came a language model able to stand alongside the best in the world, built on second-tier hardware by a group that fit in one room. And when it was ready, they published it for anyone to use. The Hangzhou Code reconstructs how that happened and, more importantly, explains why it mattered. A result like that does not simply add one more competitor: it forces a recalculation of what these systems really cost to build, how much advantage the most expensive hardware actually buys, and who owns a technology when someone decides not to charge for it. What you will find inside: • How a researcher is formed somewhere that appears on no map of the industry. • What running a quantitative fund has to do with training a language model. • Why working under constraints can produce better answers than working without them. • What changes when a capability believed to belong to a few stops belonging to them. • What is gained and what is lost by giving away something others spent fortunes on. • How to measure the true cost of an AI system, beyond the invoice. The telling is explanatory throughout: every technical idea is laid out before it is used, with enough detail to follow the decision and no assumed background in computer science. Who this book is for: readers who follow technology and want the full story behind a headline; professionals who need to understand how power is currently distributed in this sector; and anyone curious about how one decision taken far from the center ends up rewriting the rules for everybody. He did not build the largest machine. He built the one that forced everyone else to explain themselves. Start at the beginning, which is not where people usually look.
-
Autore:
-
Anno edizione:2026
-
Editore:
-
Formato:
-
Lingua:Inglese
Formato:
Gli eBook venduti da Feltrinelli.it sono in formato ePub e possono essere protetti da Adobe DRM. In caso di download di un file protetto da DRM si otterrà un file in formato .acs, (Adobe Content Server Message), che dovrà essere aperto tramite Adobe Digital Editions e autorizzato tramite un account Adobe, prima di poter essere letto su pc o trasferito su dispositivi compatibili.
Cloud:
Gli eBook venduti da Feltrinelli.it sono sincronizzati automaticamente su tutti i client di lettura Kobo successivamente all’acquisto. Grazie al Cloud Kobo i progressi di lettura, le note, le evidenziazioni vengono salvati e sincronizzati automaticamente su tutti i dispositivi e le APP di lettura Kobo utilizzati per la lettura.
Clicca qui per sapere come scaricare gli ebook utilizzando un pc con sistema operativo Windows