45 citations · 46 across the 4 of their papers we have counts for
1 paper · 1 filter
Idan Schwartz
Assessing an AI agent that can converse in human language and understand visual content is challenging. Generation metrics, such as BLEU scores favor correct syntax over semantics.…