# Latent Knowledge as a Predictor of Fact Acquisition in Fine-Tuned Large Language Models

arXiv: https://arxiv.org/abs/2601.18468 · HTML: https://mlsanity.com/paper/2601.18468/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 26 of its 49 references that are in the map.
- First posted on arXiv 26 Jan 2026.

## Builds on (most cited first, 26 of 26)

- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [Large Language Models Struggle to Learn Long-Tail Knowledge](https://mlsanity.com/paper/2211.08411/) (2022)
- [Fine-Tuning Pretrained Language Models: Weight Initializations, Data Orders, and Early Stopping](https://mlsanity.com/paper/2002.06305/) (2020)
- [Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?](https://mlsanity.com/paper/2405.05904/) (2024)
- [Survey on Factuality in Large Language Models: Knowledge, Retrieval and Domain-Specificity](https://mlsanity.com/paper/2310.07521/) (2023)
- [Fine-Tuning or Retrieval? Comparing Knowledge Injection in LLMs](https://mlsanity.com/paper/2312.05934/) (2023)
- [How Do Large Language Models Acquire Factual Knowledge During Pretraining?](https://mlsanity.com/paper/2406.11813/) (2024)
- [Injecting New Knowledge into Large Language Models via Supervised Fine-Tuning](https://mlsanity.com/paper/2404.00213/) (2024)
- [Scaling Laws for Fact Memorization of Large Language Models](https://mlsanity.com/paper/2406.15720/) (2024)
- [Understanding Finetuning for Factual Knowledge Extraction](https://mlsanity.com/paper/2406.14785/) (2024)
- [How Does Fine-tuning Affect the Geometry of Embedding Space: A Case Study on Isotropy](https://mlsanity.com/paper/2109.04740/) (2021)
- [Probing the Decision Boundaries of In-context Learning in Large Language Models](https://mlsanity.com/paper/2406.11233/) (2024)
- [Extractive Structures Learned in Pretraining Enable Generalization on Finetuned Facts](https://mlsanity.com/paper/2412.04614/) (2024)
- [How Much Knowledge Can You Pack into a LoRA Adapter without Harming LLM?](https://mlsanity.com/paper/2502.14502/) (2025)
- [FineTuneBench: How well do commercial fine-tuning APIs infuse knowledge into LLMs?](https://mlsanity.com/paper/2411.05059/) (2024)
- [On the Role of Long-tail Knowledge in Retrieval Augmented Large Language Models](https://mlsanity.com/paper/2406.16367/) (2024)
- [LLMs as Repositories of Factual Knowledge: Limitations and Solutions](https://mlsanity.com/paper/2501.12774/) (2025)
- [Understanding Memorisation in LLMs: Dynamics, Influencing Factors, and Implications](https://mlsanity.com/paper/2407.19262/) (2024)
- [Popular but Wrong: Understanding and Mitigating LLM Overconfidence through Knowledge Popularity](https://mlsanity.com/paper/2505.17537/) (2025)
- [Rote Learning Considered Useful: Generalizing over Memorized Data in LLMs](https://mlsanity.com/paper/2507.21914/) (2025)
- [Meta-Cognitive Analysis: Evaluating Declarative and Procedural Knowledge in Datasets and Large Language Models](https://mlsanity.com/paper/2403.09750/) (2024)
- [Can We Edit LLMs for Long-Tail Biomedical Knowledge?](https://mlsanity.com/paper/2504.10421/) (2025)
- [Predicting Failures of LLMs to Link Biomedical Ontology Terms to Identifiers Evidence Across Models and Ontologies](https://mlsanity.com/paper/2509.04458/) (2025)
- [Mapping Biomedical Ontology Terms to IDs: Effect of Domain Prevalence on Prediction Accuracy](https://mlsanity.com/paper/2409.13746/) (2024)
- [From Data to Knowledge: Evaluating How Efficiently Language Models Learn Facts](https://mlsanity.com/paper/2506.16912/) (2025)
- [From Memorization to Generalization: Fine-Tuning Large Language Models for Biomedical Term-to-Identifier Normalization](https://mlsanity.com/paper/2510.19036/) (2025)

## Abstract (arXiv)

Large language models store biomedical facts with uneven strength after pretraining: some facts are present in the weights but are not reliably accessible under deterministic decoding (latent knowledge), while others are scarcely represented. We fine tuned Llama 3.1 8B Instruct to learn ontology term identifier mappings from the Human Phenotype Ontology (800 pairs) and the Gene Ontology (400 training pairs), withholding 400 GO pairs to test generalization. Treating learning as a time to event process across 20 epochs, we used stochastic decoding to detect latent knowledge at baseline and Cox proportional hazards models to identify predictors of acquisition, generalization, and degradation. Baseline deterministic recall for HPO was 2.8%, rising to 71.9% after fine-tuning. Latent knowledge was the strongest predictor of faster fact acquisition (HR 2.6) and was associated with earlier, higher peak learning rates and faster convergence; identifier frequency and curated annotation counts had smaller effects. Generalization to withheld GO facts was uncommon (5.8%) but more likely when latent knowledge was present. Previously correct GO mappings degraded more often for withheld (unseen) terms than for trained (seen) terms, suggesting a protective effect of reinforcement during training. These results show that latent knowledge predicts both the speed of factual learning during fine-tuning and the limited generalization of unseen ontology facts, while resistance to degradation depends on whether facts are reinforced.