Abstract
The performance of modern language models (LMs) has been improved by chain-of-thought (CoT) reasoning, i.e., the process of generating intermediate results that guide the model towards a final answer. A possible explanation for this improvement is that CoT reasoning extends an LM’s computational power, as RNNs and transformers with additional scratch space are known to be Turing complete. Comparing LMs to Turing machines, however, introduces a category error—Turing machines decide language membership, whereas LMs define distributions over strings. To bridge this gap, we formalize CoT reasoning in a probabilistic setting. We present several results on the representational capacity of recurrent and transformer LMs with CoT reasoning, showing that they can represent the same family of distributions over strings as probabilistic Turing machines.- Anthology ID:
- 2024.acl-long.676
- Volume:
- Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- August
- Year:
- 2024
- Address:
- Bangkok, Thailand
- Editors:
- Lun-Wei Ku, Andre Martins, Vivek Srikumar
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 12510–12548
- Language:
- URL:
- https://aclanthology.org/2024.acl-long.676
- DOI:
- Cite (ACL):
- Franz Nowak, Anej Svete, Alexandra Butoi, and Ryan Cotterell. 2024. On the Representational Capacity of Neural Language Models with Chain-of-Thought Reasoning. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 12510–12548, Bangkok, Thailand. Association for Computational Linguistics.
- Cite (Informal):
- On the Representational Capacity of Neural Language Models with Chain-of-Thought Reasoning (Nowak et al., ACL 2024)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-4/2024.acl-long.676.pdf