Cover Image

Learning from Saturated Data: Signals Beyond Correctness for LLM Training

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Hiss, Hanno, Dekoninck, Jasper, Vechev, Martin
Format:	Preprint
Published:	2026
Subjects:	Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Not All Proofs Are Equal: Evaluating LLM Proof Quality Beyond Correctness
by: Petrov, Ivo, et al.
Published: (2026)

Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation
by: Dekoninck, Jasper, et al.
Published: (2024)

A Unified Approach to Routing and Cascading for LLMs
by: Dekoninck, Jasper, et al.
Published: (2024)

ConStat: Performance-Based Contamination Detection in Large Language Models
by: Dekoninck, Jasper, et al.
Published: (2024)

BrokenMath: A Benchmark for Sycophancy in Theorem Proving with LLMs
by: Petrov, Ivo, et al.
Published: (2025)

Controlled Text Generation via Language Model Arithmetic
by: Dekoninck, Jasper, et al.
Published: (2023)

Constrained Decoding of Diffusion LLMs with Context-Free Grammars
by: Mündler, Niels, et al.
Published: (2025)

Beyond Benchmarks: MathArena as an Evaluation Platform for Mathematics with LLMs
by: Dekoninck, Jasper, et al.
Published: (2026)

MathArena: Evaluating LLMs on Uncontaminated Math Competitions
by: Balunović, Mislav, et al.
Published: (2025)

Evading Data Contamination Detection for Language Models is (too) Easy
by: Dekoninck, Jasper, et al.
Published: (2024)

Proof or Bluff? Evaluating LLMs on 2025 USA Math Olympiad
by: Petrov, Ivo, et al.
Published: (2025)

Adaptive Generation of Bias-Eliciting Questions for LLMs
by: Staab, Robin, et al.
Published: (2025)

The Open Proof Corpus: A Large-Scale Study of LLM-Generated Mathematical Proofs
by: Dekoninck, Jasper, et al.
Published: (2025)

MathConstruct: Challenging LLM Reasoning with Constructive Proofs
by: Balunović, Mislav, et al.
Published: (2025)

Beyond Public Access in LLM Pre-Training Data
by: Rosenblat, Sruly, et al.
Published: (2025)

Guiding LLMs The Right Way: Fast, Non-Invasive Constrained Generation
by: Beurer-Kellner, Luca, et al.
Published: (2024)

Recovered in Translation: Efficient Pipeline for Automated Translation of Benchmarks and Datasets
by: Yukhymenko, Hanna, et al.
Published: (2026)

SEAL: Can Saturated Benchmarks Be Revived by LLM-as-a-Meta-Judge?
by: Chen, Jiamin, et al.
Published: (2026)

Beyond Two-Stage Training: Cooperative SFT and RL for LLM Reasoning
by: Chen, Liang, et al.
Published: (2025)

Metric-Dependent Annotation Saturation for Learning from Label Distributions
by: Kohli, Guneet
Published: (2026)

COMPL-AI Framework: A Technical Interpretation and LLM Benchmarking Suite for the EU Artificial Intelligence Act
by: Guldimann, Philipp, et al.
Published: (2024)

Tailored Truths: Optimizing LLM Persuasion with Personalization and Fabricated Statistics
by: Timm, Jasper, et al.
Published: (2025)

Training-Free Loosely Speculative Decoding: Accepting Semantically Correct Drafts Beyond Exact Match
by: Li, Jinze, et al.
Published: (2025)

SyncThink: A Training-Free Strategy to Align Inference Termination with Reasoning Saturation
by: Li, Gengyang, et al.
Published: (2026)

A Training-free LLM-based Approach to General Chinese Character Error Correction
by: Zhou, Houquan, et al.
Published: (2025)

A Synthetic Dataset for Personal Attribute Inference
by: Yukhymenko, Hanna, et al.
Published: (2024)

Self-contradictory Hallucinations of Large Language Models: Evaluation, Detection and Mitigation
by: Mündler, Niels, et al.
Published: (2023)

Saturation-Driven Dataset Generation for LLM Mathematical Reasoning in the TPTP Ecosystem
by: Quesnel, Valentin, et al.
Published: (2025)

The Limits of Data Scaling: Sub-token Utilization and Acoustic Saturation in Multilingual ASR
by: Liang, Siyu, et al.
Published: (2025)

CuTS: Customizable Tabular Synthetic Data Generation
by: Vero, Mark, et al.
Published: (2023)

Beyond Correctness: Learning Robust Reasoning via Transfer
by: Lee, Hyunseok, et al.
Published: (2026)

LLM Reasoning as Trajectories: Step-Specific Representation Geometry and Correctness Signals
by: Sun, Lihao, et al.
Published: (2026)

HaluNet: Learning Hallucination Risk from Internal Signals in LLM Question Answering
by: Tong, Chaodong, et al.
Published: (2025)

Learning from Mistakes: Self-correct Adversarial Training for Chinese Unnatural Text Correction
by: Feng, Xuan, et al.
Published: (2024)

Addressing Performance Saturation for LLM RL via Precise Entropy Curve Control
by: Li, Bolian, et al.
Published: (2026)

Reinforcement Learning for LLM Post-Training: A Survey
by: Wang, Zhichao, et al.
Published: (2024)

Learn Beyond The Answer: Training Language Models with Reflection for Mathematical Reasoning
by: Zhang, Zhihan, et al.
Published: (2024)

Detecting Non-Membership in LLM Training Data via Rank Correlations
by: Shetty, Pranav, et al.
Published: (2026)

Common Corpus: The Largest Collection of Ethical Data for LLM Pre-Training
by: Langlais, Pierre-Carl, et al.
Published: (2025)

Simulating Training Data Leakage in Multiple-Choice Benchmarks for LLM Evaluation
by: Hidayat, Naila Shafirni, et al.
Published: (2025)