
多言語基盤モデルの評価において、SeaEvalは言語理解、文化理解、論理推論の能力を包括的に検証する。


coremsg

seaeval-for-multilingual-foundation-models-evaluating-language-culture-and-reasoning


SeaEval for Multilingual Foundation Models: Evaluating Language, Culture, and Reasoning


title_rewrite


Large language models can be effective user simulators for conversational recommendation, but they may exhibit deviations from human behavior that can be reduced with model selection and prompting strategies.


evaluating-large-language-models-as-generative-user-simulators-for-conversational-recommendation


Evaluating Large Language Models as Generative User Simulators for Conversational Recommendation
