Pith. sign in

REVIEW 6 cited by

Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2407.14985 v5 pith:PYUHV3XS submitted 2024-07-20 cs.CL cs.AIcs.LG

Generalization v.s. Memorization: Tracing Language Models' Capabilities Back to Pretraining Data

classification cs.CL cs.AIcs.LG
keywords memorizationpretrainingtasksdatamodelsansweringfactualgeneralization
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

The impressive capabilities of large language models (LLMs) have sparked debate over whether these models genuinely generalize to unseen tasks or predominantly rely on memorizing vast amounts of pretraining data. To explore this issue, we introduce an extended concept of memorization, distributional memorization, which measures the correlation between the LLM output probabilities and the pretraining data frequency. To effectively capture task-specific pretraining data frequency, we propose a novel task-gram language model, which is built by counting the co-occurrence of semantically related $n$-gram pairs from task inputs and outputs in the pretraining corpus. Using the Pythia models trained on the Pile dataset, we evaluate four distinct tasks: machine translation, factual question answering, world knowledge understanding, and math reasoning. Our findings reveal varying levels of memorization, with the strongest effect observed in factual question answering. Furthermore, while model performance improves across all tasks as LLM size increases, only factual question answering shows an increase in memorization, whereas machine translation and reasoning tasks exhibit greater generalization, producing more novel outputs. This study demonstrates that memorization plays a larger role in simpler, knowledge-intensive tasks, while generalization is the key for harder, reasoning-based tasks, providing a scalable method for analyzing large pretraining corpora in greater depth.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 6 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. LLMs Can Leak Training Data But Do They Want To? A Propensity-Aware Evaluation of Memorization in LLMs

    cs.CL 2026-06 unverdicted novelty 7.0

    LLMs show high memorization capability under prefix attacks but low propensity under generic or dataset-specific prompts, with continual pre-training further reducing both.

  2. Access Paths for Efficient Ordering with Large Language Models

    cs.DB 2025-08 unverdicted novelty 6.0

    Introduces the LLM ORDER BY semantic operator with algorithmic improvements, a semantic-aware external merge sort, and a budget-aware optimizer that selects near-optimal access paths for LLM-based ordering.

  3. SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training

    cs.AI 2025-01 unverdicted novelty 6.0

    Reinforcement learning post-training enables generalization to unseen textual rule variants and visual changes in foundation models, while supervised fine-tuning primarily leads to memorization.

  4. Train Smarter, Not Longer: Memorization-Guided Data Reuse for Efficient LLM Training

    cs.LG 2026-07 conditional novelty 5.0

    Properly spaced multi-epoch reuse of high-quality data, guided by a measured memorization window, continues to improve LLM performance far beyond the common four-epoch heuristic.

  5. Remembering Unequally: Global and Disciplinary Bias in LLM Reconstruction of Scholarly Coauthor Lists

    cs.CL 2025-11 unverdicted novelty 5.0

    LLMs show systematic bias toward highly cited scholars when reconstructing coauthor lists, with more balanced outcomes in fields like Clinical Medicine and some African regions.

  6. Attributing Culture-Conditioned Generations to Pretraining Corpora

    cs.CL 2024-12 unverdicted novelty 5.0

    MEMOed framework attributes LLM generations about cultures to pretraining memorization and finds frequency-based biases across 110 cultures for food and clothing.