China Telecom released a single dense trillion parameter semantic model...

China Telecom semantic model

DIYuan | 2024-06-20 17:17

【數(shù)據(jù)猿導(dǎo)讀】 China Telecom released a single dense trillion parameter semantic model

On June 19, it was learned from China Telecom that China Telecom Artificial Intelligence Research Institute (TeleAI) and Beijing Zhiyuan Artificial Intelligence Research Institute released the world's first single dense trillion parameter semantic model Tele-FLM-1T, becoming the first institution to release dense trillion parameter large model in China. The reporter learned that in response to the problem of high computing power consumption in large model training, Tele-FLM series models jointly developed by TeleAI and Zhiyuan based on key technologies such as model growth and loss prediction only use 9% of the computing power resources of the industry's ordinary training programs, based on 112 A800 servers. Completed the training of 3 models with a total of 2.3T tokens in 4 months.

來源：DIYuan

收藏分享

聲明：數(shù)據(jù)猿尊重媒體行業(yè)規(guī)范，相關(guān)內(nèi)容都會(huì)注明來源與作者；轉(zhuǎn)載我們原創(chuàng)內(nèi)容時(shí)，也請務(wù)必注明“來源：數(shù)據(jù)猿”與作者名稱，否則將會(huì)受到數(shù)據(jù)猿追責(zé)。