Exploring the reversal curse and other deductive logical reasoning in BERT and GPT-based large language models
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
The "Reversal Curse" describes the inability of autoregressive decoder large language models (LLMs) to deduce "B is A" from "A is B," assuming that B and A are distinct and can be uniquely identified from each other. This logical failure suggests limitations in using generative pretrained transformer (GPT) models for tasks like constructing knowledge graphs. Our study revealed that a bidirectional LLM, bidirectional encoder representations from transformers (BERT), does not suffer from this issue. To investigate further, we focused on more complex deductive reasoning by training encoder and decoder LLMs to perform union and intersection operations on sets. While both types of models managed tasks involving two sets, they struggled with operations involving three sets. Our findings underscore the differences between encoder and decoder models in handling logical reasoning. Thus, selecting BERT or GPT should depend on the task's specific needs, utilizing BERT's bidirectional context comprehension or GPT's sequence prediction strengths.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it