1 paper · 1 filter
Jie Ma, Jie Gao, Yihang Liu +5
JEV models make multimodal decisions by directly scoring candidates. Although the common context is encoded once, candidate evaluation can still repeat matching token histories, du…