Speech-trained and conversation-fine-tuned LLMs show small, inconsistent accuracy gains over text-only models on detecting covert deception, but the comparison is confounded.
In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 289–299
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
"Yeah Right!" -- Do LLMs Exhibit Multimodal Feature Transfer?
Speech-trained and conversation-fine-tuned LLMs show small, inconsistent accuracy gains over text-only models on detecting covert deception, but the comparison is confounded.