The paper introduces Witness Overlap, a new method for determining directional provenance among open-weight model checkpoints. By using a third same-family checkpoint as a witness, the technique compares local geometry to infer which checkpoint is the parent of another, achieving 95.3% accuracy on 176 LLM checkpoints from 16 families. The authors also demonstrate robustness to weight noise and pruning, and extend the approach to VLM and diffusion families.
arXiv:2608. 14929v1 Announce Type: cross Abstract: Open-weight language models are fine-tuned, quantized, pruned, and merged, yet their provenance is often undocumented.
arXiv:2607. 10617v1 Announce Type: new Abstract: The lineage graph of open-weight language models is self-reported: Hugging Face's base_model metadata field is optional and unverified, and over 60% of Hub models document no parentage at all.
arXiv:2607. 25880v1 Announce Type: cross Abstract: LLM provenance testing asks whether a suspect LLM belongs to the same lineage as a source.
arXiv:2606. 19625v2 Announce Type: replace-cross Abstract: We use training-data attribution as an interpretable tool for capability discovery, mapping which regions of the pretraining corpus support social-reasoning versus STEM-reasoning in OLMo3-7B.
arXiv:2608. 02877v1 Announce Type: new Abstract: Causal discovery recovers directed structure from observational data and is increasingly used in clinical settings to support mechanism reasoning and fairness audits of predictive models.
arXiv:2608. 20187v1 Announce Type: cross Abstract: Practitioners inferring causality from observational data usually rely on a single method and treat its output as causal truth.
arXiv:2605. 03534v2 Announce Type: replace-cross Abstract: Retrieval-augmented generation (RAG) grounds answers in retrieved passages, yet relevance does not guarantee sufficiency: a topical passage may still fail to justify the answer.
The paper introduces the concept of an epistemic Sybil problem in multi‑agent AI systems, where multiple agents may produce seemingly independent reports that actually stem from the same underlying evidence. It formalizes this issue using information‑theoretic measures and demonstrates through large‑scale experiments that naive aggregation of replicated reports can severely degrade inference accuracy unless the system accounts for shared evidence ancestry and correlated extraction errors. The study shows that aggregators that track evidential dependence rather than merely report multiplicity or similarity achieve better calibration and inference performance.
arXiv:2610.07518v1 Announce Type: cross Abstract: Safety auditing of post-trained large language models typically relies on model behavior, requiring model execution and depending on the coverage of...
arXiv:2607. 09689v3 Announce Type: replace Abstract: Snapshot-backed sandboxes make branching cheap while leaving evidence dependence unchanged.
The paper introduces a reference‑based bias detection method that audits hidden‑state representations of language models by encoding sentences as similarities to a fixed set of anchor sentences. This relative representation allows comparison across model variants, such as before and after fine‑tuning, and yields a metric called Representational Bias Shift (ΔB). ΔB correlates strongly with output‑level bias changes, can detect bias‑increasing checkpoints with high ROC AUC, and is computationally efficient, requiring only a few minutes and far less compute than traditional benchmarks.
arXiv:2609.36569v1 Announce Type: cross Abstract: Checkpoint selection is a routine decision in supervised fine-tuning (SFT): training produces multiple checkpoints, but only one is retained. Yet fix...