Entity · paper

Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs

paperactivewhich-models-are-our-models-built-on-auditing-invisible-dependencies-in-modern-llms-8851393e·1 events·first seen Jun 11, 2026

Aliases: Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs

Co-occurring entities

ModSleuth

More like this (12)

Do Language Models Dream of Binding Molecules? Benchmarking LLMs under Spatial Constraints Co-LMLM: Continuous-Query Limited Memory Language Models Where Do Models Find Happiness? Emotion Vectors in Open-Source LLMs Operadic consistency: a label-free signal for compositional reasoning failures in LLMs Reason-Mediated Behavioral Models for Auditing LLM Social Simulators Knowledge Knows, Verbalization Tells: Disentangling Latent Directions for Mathematical Solvability in LLMs Continual LLM Upcycling: A Predictor-Gated Bank-Wise Sparsity Training Recipe for Dense-to-Sparse LLMs Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs Operads for compositional reasoning in LLMs code synthesis LLMs Surrogate Fidelity: When Can Open LLMs Explain Closed Ones?Inside the Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge Bias

Recent events (1)

6arXiv · cs.CL·Jun 11, 2026·source ↗

ModSleuth: Agentic system audits invisible dependency graphs in modern LLM training pipelines

Researchers introduce ModSleuth, an agentic system that recursively reconstructs LLM dependency graphs from public artifacts, recovering 1,060 source-verified dependencies across four major LLM releases. The system formalizes direct and indirect dependencies and operation-centered relationships to handle fragmented, inconsistent documentation. Applied at scale, the resulting graphs expose multi-hop license obligations, train-evaluation coupling, and discrepancies between released and training-time artifacts — issues that are practically invisible to manual auditing.

Evaluation and Benchmarking AI Safety Research ModSleuth Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs