Exploring the Reversal Curse and Other Deductive Logical Reasoning in BERT and GPT-Based Large Language Models

Abstract

The term "Reversal Curse" refers to the scenario where auto-regressive decoder large language models (LLMs), such as ChatGPT, trained on "A is B" fail to learn "B is A," assuming that B and A are distinct and can be uniquely identified from each other, demonstrating a basic failure of logical deduction. This raises a red flag in the use of GPT models for certain general tasks such as constructing knowledge graphs, considering their adherence to this symmetric principle. In our study, we examined a bidirectional LLM, BERT, and found that it is immune to the reversal curse. Driven by ongoing efforts to construct biomedical knowledge graphs with LLMs, we also embarked on evaluating more complex but essential deductive reasoning capabilities. This process included first training encoder and decoder language models to master the intersection and union operations on two sets and then moving on to assess their capability to infer different combinations of union and intersection operations on three newly created sets. The findings showed that while both encoder and decoder language models, trained for tasks involving two sets (union/intersection), were proficient in such scenarios, they encountered difficulties when dealing with operations that included three sets (various combinations of union and intersection). Our research highlights the distinct characteristics of encoder and decoder models in simple and complex logical reasoning. In practice, the choice between BERT and GPT should be guided by the specific requirements and nature of the task at hand, leveraging their respective strengths in bidirectional context comprehension and sequence prediction.

Final revision. To appear in Patterns

References in corpus (16)

• LLaMA: Open and Efficient Foundation Language Models
• Llama 2: Open Foundation and Fine-Tuned Chat Models
• LoRA: Low-Rank Adaptation of Large Language Models
• Structured prompt interrogation and recursive extraction of semantics (SPIRES): A method for populating knowledge bases using zero-shot learning
• Chatlaw: A Multi-Agent Legal Assistant based on a Role-Aligned Mixture-of-Experts Architecture
• Enhancing Phenotype Recognition in Clinical Notes Using Large Language Models: PhenoBCBERT and PhenoGPT
• TinyStories: How Small Can Language Models Be and Still Speak Coherent English?
• Determinants of LLM-assisted Decision-Making
• The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
• Studying Large Language Model Generalization with Influence Functions
• SocraSynth: Multi-LLM Reasoning with Conditional Statistics
• Taken out of context: On measuring situational awareness in LLMs
• Can an LLM Learn Preferences from Choice Data?
• Classification of integers based on residue classes via modern deep learning algorithms
• Experimenting with ChatGPT for Spreadsheet Formula Generation: Evidence of Risk in AI Generated Spreadsheets
• AutoRD: An Automatic and End-to-End System for Rare Disease Knowledge Graph Construction Based on Ontologies-enhanced Large Language Models

Cited by in corpus (2)

• Integrating Chain-of-Thought and Retrieval Augmented Generation Enhances Rare Disease Diagnosis from Clinical Notes
• Multimodal Integrated Knowledge Transfer to Large Language Models through Preference Optimization with Biomedical Applications