Yeqa uye kokuyinhloko
Ama-modeli kanye nobuchwepheshe

I-Meta ivula umthombo wemodeli ye-AI engu-30B i-Muse Glimmer, i-quantization engu-4bit isebenza ku-GPU eyodwa ene-24GB VRAM

I-Meta namuhla ikhiphe imodeli ye-AI i-Muse Glimmer, futhi yavula umthombo wezisindo zale modeli ngaphansi kwelayisensi ye-Apache 2.0.

IT ikhayaCishe imizuzu engu-4 yokufunda
I-Meta ivula umthombo wemodeli ye-AI engu-30B i-Muse Glimmer, i-quantization engu-4bit isebenza ku-GPU eyodwa ene-24GB VRAM

I-Meta namuhla ikhiphe imodeli ye-AI i-Muse Glimmer, futhi yavula umthombo wezisindo zale modeli ngaphansi kwelayisensi ye-Apache 2.0. I-Muse Glimmer iyimodeli enamapharamitha ayizigidi eziyizinkulungwane ezingu-30 (30B), eyakhelwe futhi yalungiselelwa ngokujulile ukuqhuba izinqubo zomsebenzi zama-agent asebenza endaweni yangakini, ahlala esebenza usuku lonke.

Inkampani ithi i-Muse Glimmer isebenzisa ubuchwepheshe be-quantization engu-4bit, obuyenza ikwazi ukusebenza endaweni yangakini ku-Mac noma ku-PC ene-24GB/32GB VRAM. Ukuhlolwa kwangempela kuqinisekisile ukuthi lokhu kucindezelwa kunomthelela omncane kakhulu ekusebenzeni kwemisebenzi yama-agent, cishe ngaphandle kokwehla nhlobo.

I-Meta ihlole le modeli ku-MacBook M4 Max, M5 Max kanye naku-RTX 5090. Imiphumela ikhombisa ukuthi ijubane layo liyakwazi ukusekela izingxoxo ezibushelelezi kanye nokuxhumana kwama-agent ngesikhathi sangempela, futhi zonke izibalo zenziwa ngokuphelele, ngo-100%, kudivayisi yasendaweni.

I-Meta ivula umthombo wemodeli ye-AI engu-30B i-Muse Glimmer, i-quantization engu-4bit isebenza ku-GPU eyodwa ene-24GB VRAM

Uma iqhathaniswa ne-Gemma4-31B kanye ne-Qwen3.6-27B, i-Muse Glimmer ikhombise ukusebenza okuqinile kumamodeli anosayizi ofanayo ezivivinyweni eziningi ezijwayelekile zokulinganisa amamodeli amakhulu olimi.

I-Meta ivula umthombo wemodeli ye-AI engu-30B i-Muse Glimmer, i-quantization engu-4bit isebenza ku-GPU eyodwa ene-24GB VRAM

Isixhumanisi somthombo ovulekile:

https://huggingface.co/meta-models/Muse-Glimmer-30B