Towards Logically Consistent Language Models via Probabilistic Reasoning (2404.12843v1)
Abstract: LLMs are a promising venue for natural language understanding and generation tasks. However, current LLMs are far from reliable: they are prone to generate non-factual information and, more crucially, to contradict themselves when prompted to reason about beliefs of the world. These problems are currently addressed with large scale fine-tuning or by delegating consistent reasoning to external tools. In this work, we strive for a middle ground and introduce a training objective based on principled probabilistic reasoning that teaches a LLM to be consistent with external knowledge in the form of a set of facts and rules. Fine-tuning with our loss on a limited set of facts enables our LLMs to be more logically consistent than previous baselines and allows them to extrapolate to unseen but semantically similar factual knowledge more systematically.
- Semantic Probabilistic Layers for Neuro-Symbolic Learning. In NeurIPS, 2022.
- Semantic sensitivities and inconsistent predictions: Measuring the fragility of nli models, 2024.
- Roberto Battiti. Maximum satisfiability problemMaximum Satisfiability Problem, pp. 2035–2041. Springer US, Boston, MA, 2009. ISBN 978-0-387-74759-0. doi: 10.1007/978-0-387-74759-0˙364.
- Discovering latent knowledge in language models without supervision, 2022.
- From statistical relational to neural-symbolic artificial intelligence. In Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence, pp. 4943–4950, 2021.
- Truthful ai: Developing and governing ai that does not lie, 2021.
- Maieutic prompting: Logically consistent reasoning with recursive explanations, 2022.
- Beliefbank: Adding memory to a pre-trained language model for a systematic notion of belief, 2021.
- Language models with rationality, 2023.
- A logic-driven framework for consistency of neural models, 2019.
- Truthfulqa: Measuring how models mimic human falsehoods, 2022.
- Vera: A general-purpose plausibility estimation model for commonsense statements, 2023.
- Roberta: A robustly optimized bert pretraining approach, 2019.
- Sgdr: Stochastic gradient descent with warm restarts. In International Conference on Learning Representations, 2016.
- Enhancing self-consistency and performance of pre-trained language models through natural language inference, 2022.
- Language models as knowledge bases?, 2019.
- Differentiation of blackbox combinatorial solvers. In 8th International Conference on Learning Representations, ICLR 2020, Addis Ababa, Ethiopia, April 26-30, 2020. OpenReview.net, 2020.
- General-purpose question-answering with macaw, 2021.
- Entailer: Answering questions with faithful and truthful chains of reasoning, 2022.
- A compositional atlas of tractable circuit operations for probabilistic inference. Advances in Neural Information Processing Systems, 34:13189–13201, 2021.
- Fact or fiction: Verifying scientific claims, 2020.
- A semantic loss function for deep learning with symbolic knowledge, 2018.
- Improved logical reasoning of language models via differentiable symbolic programming, 2023.
Collections
Sign up for free to add this paper to one or more collections.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.