Complexity Control Facilitates Reasoning-Based Compositional Generalization in Transformers
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866912189548331008 |
|---|---|
| author | Zhang, Zhongwang Lin, Pengxiao Wang, Zhiwei Zhang, Yaoyu Xu, Zhi-Qin John |
| author_facet | Zhang, Zhongwang Lin, Pengxiao Wang, Zhiwei Zhang, Yaoyu Xu, Zhi-Qin John |
| contents | Transformers have demonstrated impressive capabilities across various tasks, yet their performance on compositional problems remains a subject of debate. In this study, we investigate the internal mechanisms underlying Transformers' behavior in compositional tasks. We find that complexity control strategies significantly influence whether the model learns primitive-level rules that generalize out-of-distribution (reasoning-based solutions) or relies solely on memorized mappings (memory-based solutions). By applying masking strategies to the model's information circuits and employing multiple complexity metrics, we reveal distinct internal working mechanisms associated with different solution types. Further analysis reveals that reasoning-based solutions exhibit a lower complexity bias, which aligns with the well-studied neuron condensation phenomenon. This lower complexity bias is hypothesized to be the key factor enabling these solutions to learn reasoning rules. We validate these conclusions across multiple real-world datasets, including image generation and natural language processing tasks, confirming the broad applicability of our findings. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_08537 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Complexity Control Facilitates Reasoning-Based Compositional Generalization in Transformers Zhang, Zhongwang Lin, Pengxiao Wang, Zhiwei Zhang, Yaoyu Xu, Zhi-Qin John Computation and Language Machine Learning Transformers have demonstrated impressive capabilities across various tasks, yet their performance on compositional problems remains a subject of debate. In this study, we investigate the internal mechanisms underlying Transformers' behavior in compositional tasks. We find that complexity control strategies significantly influence whether the model learns primitive-level rules that generalize out-of-distribution (reasoning-based solutions) or relies solely on memorized mappings (memory-based solutions). By applying masking strategies to the model's information circuits and employing multiple complexity metrics, we reveal distinct internal working mechanisms associated with different solution types. Further analysis reveals that reasoning-based solutions exhibit a lower complexity bias, which aligns with the well-studied neuron condensation phenomenon. This lower complexity bias is hypothesized to be the key factor enabling these solutions to learn reasoning rules. We validate these conclusions across multiple real-world datasets, including image generation and natural language processing tasks, confirming the broad applicability of our findings. |
| title | Complexity Control Facilitates Reasoning-Based Compositional Generalization in Transformers |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2501.08537 |