Understanding Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation

Xinyi Wang, Alfonso Amayuelas, Kexun Zhang, Liangming Pan, Wenhu Chen, William Yang Wang
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:50026-50042, 2024.

Abstract

Pre-trained language models (LMs) are able to perform complex reasoning without explicit fine-tuning. To understand how pre-training with a next-token prediction objective contributes to the emergence of such reasoning capability, we propose that we can view an LM as deriving new conclusions by aggregating indirect reasoning paths seen at pre-training time. We found this perspective effective in two important cases of reasoning: logic reasoning with knowledge graphs (KGs) and chain-of-thought (CoT) reasoning. More specifically, we formalize the reasoning paths as random walk paths on the knowledge/reasoning graphs. Analyses of learned LM distributions suggest that a weighted sum of relevant random walk path probabilities is a reasonable way to explain how LMs reason. Experiments and analysis on multiple KG and CoT datasets reveal the effect of training on random walk paths and suggest that augmenting unlabeled random walk reasoning paths can improve real-world multi-step reasoning performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-wang24a, title = {Understanding Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation}, author = {Wang, Xinyi and Amayuelas, Alfonso and Zhang, Kexun and Pan, Liangming and Chen, Wenhu and Wang, William Yang}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {50026--50042}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/wang24a/wang24a.pdf}, url = {https://proceedings.mlr.press/v235/wang24a.html}, abstract = {Pre-trained language models (LMs) are able to perform complex reasoning without explicit fine-tuning. To understand how pre-training with a next-token prediction objective contributes to the emergence of such reasoning capability, we propose that we can view an LM as deriving new conclusions by aggregating indirect reasoning paths seen at pre-training time. We found this perspective effective in two important cases of reasoning: logic reasoning with knowledge graphs (KGs) and chain-of-thought (CoT) reasoning. More specifically, we formalize the reasoning paths as random walk paths on the knowledge/reasoning graphs. Analyses of learned LM distributions suggest that a weighted sum of relevant random walk path probabilities is a reasonable way to explain how LMs reason. Experiments and analysis on multiple KG and CoT datasets reveal the effect of training on random walk paths and suggest that augmenting unlabeled random walk reasoning paths can improve real-world multi-step reasoning performance.} }
Endnote
%0 Conference Paper %T Understanding Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation %A Xinyi Wang %A Alfonso Amayuelas %A Kexun Zhang %A Liangming Pan %A Wenhu Chen %A William Yang Wang %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-wang24a %I PMLR %P 50026--50042 %U https://proceedings.mlr.press/v235/wang24a.html %V 235 %X Pre-trained language models (LMs) are able to perform complex reasoning without explicit fine-tuning. To understand how pre-training with a next-token prediction objective contributes to the emergence of such reasoning capability, we propose that we can view an LM as deriving new conclusions by aggregating indirect reasoning paths seen at pre-training time. We found this perspective effective in two important cases of reasoning: logic reasoning with knowledge graphs (KGs) and chain-of-thought (CoT) reasoning. More specifically, we formalize the reasoning paths as random walk paths on the knowledge/reasoning graphs. Analyses of learned LM distributions suggest that a weighted sum of relevant random walk path probabilities is a reasonable way to explain how LMs reason. Experiments and analysis on multiple KG and CoT datasets reveal the effect of training on random walk paths and suggest that augmenting unlabeled random walk reasoning paths can improve real-world multi-step reasoning performance.
APA
Wang, X., Amayuelas, A., Zhang, K., Pan, L., Chen, W. & Wang, W.Y.. (2024). Understanding Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:50026-50042 Available from https://proceedings.mlr.press/v235/wang24a.html.

Related Material