Pith. sign in

REVIEW 15 cited by

On Memorization of Large Language Models in Logical Reasoning

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 2410.23123 v2 pith:W3AK36ZL submitted 2024-10-30 cs.CL

On Memorization of Large Language Models in Logical Reasoning

classification cs.CL
keywords reasoningmemorizationllmspuzzlesfine-tuninglogicalperformancebenchmarks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Large language models (LLMs) achieve good performance on challenging reasoning benchmarks, yet could also make basic reasoning mistakes. This contrasting behavior is puzzling when it comes to understanding the mechanisms behind LLMs' reasoning capabilities. One hypothesis is that the increasingly high and nearly saturated performance on common reasoning benchmarks could be due to the memorization of similar problems. In this paper, we systematically investigate this hypothesis with a quantitative measurement of memorization in reasoning tasks, using a dynamically generated logical reasoning benchmark based on Knights and Knaves (K&K) puzzles. We find that LLMs could interpolate and memorize the training puzzles (achieving near-perfect accuracy) after fine-tuning, yet they struggle with slight variations of these puzzles. On the other hand, we show that while fine-tuning leads to heavy memorization, it also consistently improves generalization performance. Through in-depth analyses with perturbation tests, cross difficulty-level transferability, probing model internals, and fine-tuning with wrong answers, we establish that LLMs develop reasoning skills on K&K puzzles alongside memorization. Finally, our analysis based on a per-sample memorization score sheds light on how LLMs switch between reasoning and memorization when solving logical puzzles. Our code and data are available at https://memkklogic.github.io.

discussion (0)

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

Forward citations

Cited by 15 Pith papers

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

  1. Unsteady Metrics and Benchmarking Cultures of AI Model Builders

    cs.AI 2026-05 accept novelty 8.0

    AI model builders mostly highlight unique benchmarks that act as flexible narrative tools for market positioning rather than standardized scientific measurements.

  2. Copyright Protection for Large Language Models: A Survey of Methods, Challenges, and Trends

    cs.CR 2025-08 accept novelty 7.0

    A survey of LLM copyright protection that unifies text watermarking, model watermarking, and model fingerprinting while presenting new coverage of fingerprint transfer and removal.

  3. Blending Supervised and Reinforcement Fine-Tuning with Prefix Sampling

    cs.LG 2025-07 unverdicted novelty 7.0

    Prefix-RFT blends SFT and RFT via prefix sampling from demonstrations to outperform standalone SFT, RFT, and mixed-policy baselines on math reasoning problems.

  4. Enough is as good as a feast: A Comprehensive Analysis of How Reinforcement Learning Mitigates Task Conflicts in LLMs

    cs.CL 2026-07 conditional novelty 6.0

    RL-trained LLMs keep most of their skills after weight merging, while SFT-trained LLMs drop about 19% on average, because RL keeps parameter updates smaller and more task-compatible.

  5. A Close Look At World Model Recovery In Supervised Fine-Tuned LLM Planners

    cs.LG 2026-06 unverdicted novelty 6.0

    Supervised fine-tuning lets LLMs linearly encode action validity and state predicates, with broader state-space coverage during training improving world-model recovery.

  6. RLVE: Scaling Up Reinforcement Learning for Language Models with Adaptive Verifiable Environments

    cs.CL 2025-11 conditional novelty 6.0

    Training a 1.5B language model on 400 adaptively-difficult, procedurally generated, verifiable environments improved average reasoning-benchmark performance by 3.37% absolute, compared with 0.49% from continuing origi...

  7. ActivationReasoning: Logical Reasoning in Latent Activation Spaces

    cs.LG 2025-10 unverdicted novelty 6.0

    ActivationReasoning grounds logical reasoning in LLM latent activations via SAEs to enable structured inference, concept composition, and behavior steering on multi-hop, abstraction, and safety tasks.

  8. Understanding the Ability of LLMs to Handle Character-Level Perturbation

    cs.CL 2025-10 conditional novelty 6.0

    LLMs remain surprisingly accurate on math and coding when invisible Unicode noise is inserted after every character, with robustness driven by implicit internal denoising and, for some models, explicit rewriting in ch...

  9. Structured In-context Environment Scaling for Large Language Model Reasoning

    cs.CL 2025-09 conditional novelty 6.0

    SIE framework automatically constructs scalable, verifiable reasoning environments from structured data, improving in-domain performance and enabling generalization to out-of-domain math and logic tasks.

  10. Learning to Refine: Self-Refinement of Parallel Reasoning in LLMs

    cs.LG 2025-08 conditional novelty 6.0

    GSR jointly trains LLMs to generate candidate solutions and refine a superior final answer from them, achieving state-of-the-art performance on five mathematical benchmarks while transferring across model scales.

  11. On the Generalization Gap in Self-Evolving Language Model Reasoning

    cs.CL 2026-05 unverdicted novelty 5.0

    Closed-loop self-evolution on LLMs improves reasoning on Knights and Knaves tasks but plateaus short of oracle-supervised levels, with multi-turn revision nearly matching it for large models.

  12. Proximal Supervised Fine-Tuning

    cs.LG 2025-08 unverdicted novelty 5.0

    PSFT modifies supervised fine-tuning by incorporating trust-region ideas from RL to constrain policy changes, yielding better out-of-domain generalization in math and human-value tasks without entropy collapse.

  13. Entropy-KL Divergence-based Token Masking: A Novel Approach for Selective Fine-tuning of Large Language Models

    cs.AI 2026-05 unverdicted novelty 4.0

    EKSFT masks high-entropy or high-KL tokens in low-data SFT to preserve pre-trained distribution and improve downstream RL performance on math reasoning tasks.

  14. Effects of Cross-lingual Evidence in Multilingual Medical Question Answering

    cs.CL 2026-04 unverdicted novelty 4.0

    Combining English and target-language web retrieval boosts medical QA for low-resource languages to match high-resource performance, while English web data benefits high-resource languages most and specialized sources...

  15. Sharpness-Guided Group Relative Policy Optimization via Probability Shaping

    cs.LG 2025-10 unverdicted novelty 4.0

    GRPO-SG is a sharpness-guided token-weighted variant of GRPO that downweights high-gradient tokens to stabilize optimization and improve generalization in reinforcement learning with verifiable rewards.