Log Probabilities Are a Reliable Estimate of Semantic Plausibility in Base and Instruction-Tuned Language Models

Log Probabilities Are a Reliable Estimate of Semantic Plausibility in Base and Instruction-Tuned Language Models Carina Kauf author Emmanuele Chersoni author Alessandro Lenci author Evelina Fedorenko author Anna A Ivanova author 2024-11 text Proceedings of the 7th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP Yonatan Belinkov editor Najoung Kim editor Jaap Jumelet editor Hosein Mohebbi editor Aaron Mueller editor Hanjie Chen editor Association for Computational Linguistics Miami, Florida, US conference publication kauf-etal-2024-log 10.18653/v1/2024.blackboxnlp-1.18 https://aclanthology.org/2024.blackboxnlp-1.18/ 2024-11 263 277