A Hopfieldian View-based Interpretation for Chain-of-Thought Reasoning
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866913394389417984 |
|---|---|
| author | Hu, Lijie Liu, Liang Yang, Shu Chen, Xin Xiao, Hongru Li, Mengdi Zhou, Pan Ali, Muhammad Asif Wang, Di |
| author_facet | Hu, Lijie Liu, Liang Yang, Shu Chen, Xin Xiao, Hongru Li, Mengdi Zhou, Pan Ali, Muhammad Asif Wang, Di |
| contents | Chain-of-Thought (CoT) holds a significant place in augmenting the reasoning performance for large language models (LLMs). While some studies focus on improving CoT accuracy through methods like retrieval enhancement, yet a rigorous explanation for why CoT achieves such success remains unclear. In this paper, we analyze CoT methods under two different settings by asking the following questions: (1) For zero-shot CoT, why does prompting the model with "let's think step by step" significantly impact its outputs? (2) For few-shot CoT, why does providing examples before questioning the model could substantially improve its reasoning ability? To answer these questions, we conduct a top-down explainable analysis from the Hopfieldian view and propose a Read-and-Control approach for controlling the accuracy of CoT. Through extensive experiments on seven datasets for three different tasks, we demonstrate that our framework can decipher the inner workings of CoT, provide reasoning error localization, and control to come up with the correct reasoning path. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_12255 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Hopfieldian View-based Interpretation for Chain-of-Thought Reasoning Hu, Lijie Liu, Liang Yang, Shu Chen, Xin Xiao, Hongru Li, Mengdi Zhou, Pan Ali, Muhammad Asif Wang, Di Computation and Language Artificial Intelligence Human-Computer Interaction Machine Learning Chain-of-Thought (CoT) holds a significant place in augmenting the reasoning performance for large language models (LLMs). While some studies focus on improving CoT accuracy through methods like retrieval enhancement, yet a rigorous explanation for why CoT achieves such success remains unclear. In this paper, we analyze CoT methods under two different settings by asking the following questions: (1) For zero-shot CoT, why does prompting the model with "let's think step by step" significantly impact its outputs? (2) For few-shot CoT, why does providing examples before questioning the model could substantially improve its reasoning ability? To answer these questions, we conduct a top-down explainable analysis from the Hopfieldian view and propose a Read-and-Control approach for controlling the accuracy of CoT. Through extensive experiments on seven datasets for three different tasks, we demonstrate that our framework can decipher the inner workings of CoT, provide reasoning error localization, and control to come up with the correct reasoning path. |
| title | A Hopfieldian View-based Interpretation for Chain-of-Thought Reasoning |
| topic | Computation and Language Artificial Intelligence Human-Computer Interaction Machine Learning |
| url | https://arxiv.org/abs/2406.12255 |