Showing cs.CLShow all
2 papers · 1 filter
cs.CL2025
The Transfer Neurons Hypothesis: An Underlying Mechanism for Language Latent Space Transitions in Multilingual LLMs
Hinata Tezuka, Naoya Inoue
Recent studies have suggested a processing framework for multilingual inputs in decoder-based LLMs: early layers convert inputs into English-centric and language-agnostic represent…
cs.CL2023
Discovering Highly Influential Shortcut Reasoning: An Automated Template-Free Approach
Daichi Haraguchi, Kiyoaki Shirai, Naoya Inoue +1
Shortcut reasoning is an irrational process of inference, which degrades the robustness of an NLP model. While a number of previous work has tackled the identification of shortcut…