Wise Source Research Institute Launches Flag Eval "Scales" Big Model Assessment System...

Scales Big Model Assessment System

DIYuan | 2023-06-09 18:34

【數(shù)據(jù)猿導(dǎo)讀】 Wise Source Research Institute Launches FlagEval "Scales" Big Model Assessment System

June 9 morning news, 2023 Beijing wisdom source conference, wisdom source research institute director Huang Tiejun announced the launch of FlagEval (scales) large language model evaluation system, aimed at "ability, task, indicators" from the three-dimensional evaluation perspective, more than 600 dimensions of the large model for a comprehensive evaluation, to establish a scientific, fair and comprehensive The system aims to establish a scientific, fair and comprehensive technical evaluation system for the Big Model. According to the introduction, the task dimension of the big model currently includes 22 subjective and objective assessment data sets, with as many as 84,433 assessment questions. Currently exploring the use of artificial intelligence technology for scientific evaluation, and strive to reduce more subjective evaluation. It is also exploring the use of large model evaluation to assist in large model pre-training.

來(lái)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(lái)源與作者；轉(zhuǎn)載我們?cè)瓌?chuàng)內(nèi)容時(shí)，也請(qǐng)務(wù)必注明“來(lái)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)。