Abaphuhlisi abavela kwiMozilla babonise ukhupho lokuqala lwento eluncedo yellamafile, ekuvumela ukuba wenze iifayile eziphunyezwayo jikelele zokusebenzisa iimodeli zolwimi lokufunda ngomatshini omkhulu (LLM). Ukusebenzisa i-llamafile, unokuthatha ifayile yeparameter yokufunda yomatshini kwifomathi yeGGUF kwaye uyiguqule ibe yifayile ephunyezwayo enokuqhuba kwiinkqubo ezintandathu ezahlukeneyo zokusebenza kwi-AMD64 kunye ne-ARM64 hardware. I-toolkit ibhalwe kwi-C / C ++ kwaye isasazwe phantsi kwelayisensi ye-Apache 2.0.
Ukwenza iimodeli, isakhelo se-llama.cpp sisetyenziselwa ukuphunyezwa kwemodeli ye-LLaMA kwi-C / C ++, ehlanganiswe kwifayile ephunyeziweyo jikelele kwifomathi ye-APE (Enyanisiweyo ephathekayo eSebenzayo), idibaniswe nefayile yomzekelo. Ikhowudi ehlanganisiweyo idityaniswe kwithala leencwadi eliqhelekileyo le-Cosmopolitan C, okwenza kube lula ukwenza ulwakhiwo lwezicelo olusebenza kwiLinux, FreeBSD, macOS, OpenBSD, NetBSD kunye neWindows. Ikhowudi igcinwe kwiinguqulelo ze-AMD64 kunye ne-ARM64 ye-architectures, kwaye iquka amacandelo okukhuphela izibalo kwi-NVIDIA GPUs (usebenzisa i-CUDA).
umthombo: opennet.ru
