Entity · technique

Factual Recall Scaling Law

techniqueactivefactual-recall-scaling-law-4ff674ad·1 events·first seen May 19, 2026

Aliases: Factual Recall Scaling Law

Co-occurring entities

Automated Reference Verification System Superposition Model (neural networks)Superposition-Inspired Signal-to-Noise Account

More like this (12)

AI scaling laws cyberattack scaling law Scaling Laws for Neural Language Models Shannon Scaling Law power-law scaling Parametric Memory Law Scaling Laws for Reward Model Overoptimization Factual Density (FD*)FairScale inference-time compute scaling Conditional Scale Entropy Responsible Scaling Policy

Recent events (1)

7arXiv · cs.CL·May 19, 2026·source ↗

Predictable Confabulations: Factual Recall by LLMs Scales with Model Size and Topic Frequency

This paper establishes a quantitative scaling law linking LLM factual recall to both model parameter count and topic frequency in training data, evaluated across 38 models on 8,900+ scholarly references. Recall quality follows a sigmoid function in the log-linear combination of these two variables, explaining 60% of variance across 16 dense models from four families and 74-94% within individual families. The authors propose a superposition-inspired mechanism where recall is gated by a signal-to-noise ratio: concept frequency provides signal and model capacity sets the noise floor. This provides a predictive framework for understanding and anticipating LLM confabulation patterns.

Frontier Model Releases Evaluation and Benchmarking Automated Reference Verification System Factual Recall Scaling Law Superposition Model (neural networks)+2 more