HejuruUbuyobozi bukurikira
Guhindura Ibishushanyo Kuri Kode Model
Tekiniki
Ubuyobozi bwa tekiniki
BentoML ni isoko ifunguye isoko ya Python ipakira imashini yiga imashini yatojwe muburyo busanzwe, bushobora kwitwa 'Bentos'.
Umva iki gitaboAI-byavuzwe · ~2 min
It bridges the gap between a model sitting in a notebook and a production service that can actually serve predictions over an API.
Iyo umuhanga mu makuru arangije amahugurwa yicyitegererezo, kuyinjiza mubikorwa mubisanzwe bisobanura intoki kwandika code ya serivise, guhuza ibishingirwaho, kubaka ishusho ya Docker, no gushakisha API. BentoML itangiza ibi. Uzigama icyitegererezo mububiko bwaho bwibanze, hanyuma usobanure urwego rwa Service hamwe na API iherezo ryarimbishijwe kugirango ukemure umwanzuro. 'Bentoml kubaka' amabwiriza apakira icyitegererezo, code yawe ya Python, verisiyo yo kwishingikirizaho, hamwe nigihe cyo gukora muburyo bwonyine, bwahinduwe na Bento. Kuva aho, 'bentoml kontineri' itanga ishusho ya OCI Docker. BentoML ishyigikira hafi ya buri cyiciro (PyTorch, TensorFlow, scikit-yige, XGBoost, Hugging Face Transformers, ONNX) kandi ikongeramo micro-batch adaptive, ihuza amatsinda yinjira mu buryo bwikora kugirango yongere GPU idahinduye code yawe.
Itsinda-ryerekana uburiganya ryabitse moderi ya XGBoost mububiko bwa BentoML kandi yubaka Bento igaragaza a / guhanura REST iherezo rya serivisi yo kwishyura guhamagara mugihe nyacyo.
Itsinda rya ML rikoresha 'bentoml kontineri' kugirango ihindure imiterere yimyumvire ya Hugging Face mumashusho ya Docker yohereza muri cluster yabo ya Kubernetes.
Gutangira bikora moderi ya Llama itunganijwe neza hamwe na OpenLLM (yubatswe kuri BentoML), yerekana ibimenyetso kuri chat UI hamwe noguhuza imihindagurikire y'ikirere bituma GPU yuzura.
Isosiyete ikora mudasobwa yapakiye amashusho ya PyTorch itondekanya hamwe numuyoboro wacyo utunganya mbere muri Bento imwe rero impinduka nyazo zikoreshwa mubwato bwamahugurwa hamwe nicyitegererezo.
BentoML yishimangiye cyane muburyo bw'ururimi runini hamwe na AI itanga serivisi, hamwe na OpenLLM na BentoCloud itanga ibisubizo byerekana ibimenyetso, autoscaling, hamwe na gahunda ya GPU. Tegereza kwishyira hamwe hamwe na optimizers nka vLLM na TensorRT-LLM, inkunga nziza ya sisitemu yo mu bwoko bwa AI igizwe na moderi nyinshi, hamwe n'inzira yoroshye kuva Bento ipakiye kugeza kuri GPU idafite seriveri. Mugihe amakipe yimutse avuye kumurongo umwe yerekeza kumurongo wa agentic, BentoML yihagararaho nkibipakira hamwe na serivise itanga ibyo bice hamwe.
Ibaruwa y'amakuru y'ubuntu
Inkuru eshatu zizewe za AI buri gitondo cy'iminsi y'akazi, zanditse mu rurimi rworoshye. Ubuntu iteka ryose, nta kwamamaza.
Imeri imwe buri munsi y'akazi. Kwiyimura mu buryo bumwe. Ntitwigera tugurisha cyangwa ngo dusangire aderesi yawe.
Isuzume
Ibisubizo ako kanya kuri buri kibazo, n'icyemezo gishobora gusangirwa gifite ID yagenzurwa umaze gutsinda isomo.
Fasha uburezi bwa AI bw'ubuntu. AI Understanding ni umuryango utegamiye kuri leta wa 501(c)(3) — nta kwamamaza, nta kwishyura, iteka ryose. Tanga impano
BentoML ni isoko ifunguye isoko ya Python ipakira imashini yiga imashini yatojwe muburyo busanzwe, bushobora kwitwa 'Bentos'. Ikemura icyuho kiri hagati yicyitegererezo cyicaye mu ikaye na serivisi itanga umusaruro ushobora gutanga ubuhanuzi kuri API.
BentoML ipakira icyitegererezo, code yayo, kwishingikirizaho, hamwe nigihe cyo gukora mugihe cyahinduwe, cyigenga cyiswe Bento.
Amatsinda yo guhuza imihindagurikire y'ikirere ibyifuzo byinjira mugihe cyogukomeza kugirango GPU ihuze kandi yinjize byinshi byinjira nta kode ihinduka.
BentoML documentation describes Runners as the legacy 1.1 computation units that could execute separately; newer BentoML versions streamline services.
'bentoml kontineri' itanga ishusho isanzwe ya OCI / Docker kuva muri Bento kugirango yoherezwe.
Gushushanya no gushiramo ibidukikije bituma serivisi ipakiwe yororoka kandi igahinduka aho ikorera hose.
Komeza wige
Abandi bayobozi batoranijwe kuriyi ngingo
HejuruUbuyobozi bukurikira
Guhindura Ibishushanyo Kuri Kode Model
Tekiniki