Papers by Zhenping Xie
Constructing Word-Context-Coupled Space Aligned with Associative Knowledge Relations for Interpretable Language Modeling (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to train language models have limitations in interpretability . a Word-Context-Coupled Space (W2CSpace) is proposed to improve the performance of pre-trained models . |
| Approach: | They propose a Word-Context-Coupled Space to replace pre-trained models with interpretable statistical logic. |
| Outcome: | The proposed language model can achieve better performance and highly credible interpretability compared to state-of-the-art methods. |