arXiv Computation and Language
Sep 28
Where a Model Sends Its Own Repeated Token
arXiv:2609. 31181v1 Announce Type: new Abstract: Black-box model identification works by scoring a model's response to natural-language prompts.
Recovery Rates Are Not Comparable Across Transcription Factors: Chance Correction for Attribution Evaluation
Related stories
arXiv Machine Learning
Oct 9
Predictive Multiplicity in Cell-Fate Assignment: Label-Free Rashomon Sets and the Limits of Per-Cell Certification
The paper introduces FateMultiplicity, a label‑free framework that builds a Rashomon set of statistically admissible models for single‑cell trajectory inference by cross‑fitting held‑out genes and non‑inferiority testing. It shows that model multiplicity is large and depends more on the diversity of the model space than its size, and that per‑cell certified fate margins do not improve assignment reliability compared to baseline models. The study also identifies two robust constructions— a margin‑erosion ratio for branch‑point detection and a certification filter that reduces disagreement with clonal outcomes— that merit further investigation.
arXiv AI
Oct 9
Unlocking the Regulatory Genome by ARGUS: An Evidence-Constrained Agentic Framework for Interpreting Single Nucleotide Variants
ARGUS is an evidence‑constrained framework that separates deterministic biological computation from large‑language‑model reasoning to interpret single nucleotide variants in noncoding regulatory regions. It uses 458 DNABERT‑based transcription factor binding models within a hypothesis‑directed loop where a planner selects evidence sources, a verifier interprets observations, and the investigation path adapts to uncertainty. In a case study of variant rs6983267, ARGUS generated divergent investigative trajectories for four transcription factors, demonstrating how real experimental data from ADASTRA, JASPAR, and ENCODE cCRE guide conclusions and how the LLM planner can reduce unnecessary tool calls.
arXiv Machine Learning
Aug 18
Training Leaves Traces: Centered Residual Signatures for Language Model Lineage Verification
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 AI
Oct 31
GenoBERT: A Language Model for Accurate Genotype Imputation
GenoBERT is a transformer-based, reference‑free model that tokenizes phased genotypes and uses self‑attention to capture short‑ and long‑range linkage disequilibrium. Benchmarking on the Louisiana Osteoporosis Study and the 1000 Genomes Project shows GenoBERT outperforms four baseline methods, achieving an overall imputation accuracy of about 0.98 at up to 25% missing data and maintaining r² > 0.90 even at 50% missingness across multiple ancestries. The model’s 128‑SNP context window is sufficient to capture local correlation structures, and it remains robust to small sample sizes and weak LD, offering a scalable solution for genotype imputation without relying on reference panels.
arXiv AI
Aug 11
TokenPrint: A Calibrated Token-Space Fingerprint for Language-Model Provenance
arXiv:2608. 08139v1 Announce Type: new Abstract: Establishing the provenance of a language model---including its base checkpoint and possible overlap in training distributions---is a governance challenge that metadata alone cannot resolve.
arXiv AI
Jul 3
Breaking Safety at the Token Boundary: How BPE Tokenization Creates Exploitable Gaps in LLM Alignment
arXiv:2607. 01239v1 Announce Type: cross Abstract: Character-level perturbations bypass safety alignment in modern LLMs despite leaving prompts human-readable.
arXiv AI
Sep 16
TAME: Token Attribution and Masking for Emergent misalignment
TAME (Token Attribution and Masking for Emergent misalignment) is a three‑stage framework that identifies which training tokens drive harmful behavior in fine‑tuned language models. It first scores tokens by how much fine‑tuning increases their likelihood, then characterizes patterns among high‑attribution tokens, and finally validates them by masking during training. Experiments on Llama and Qwen show that masking the top‑attribution tokens reduces emergent misalignment by 23‑ to 36‑fold, while random masking has no effect.
arXiv AI
Jul 23
Causal dictionary learning reveals and validates transcription-factor binding features in genomic language models
arXiv:2607. 19618v1 Announce Type: cross Abstract: Genomic language models achieve strong performance across regulatory-genomics tasks, yet what these models internally represent remains opaque, and the field lacks a principled procedure for verifying that an apparent ``concept'' inside a model is real rather than an artifact of sequence composition.
arXiv Machine Learning
Sep 11
A Fragility Spectrum for Recursive Language-Model Training
The paper investigates how recursive contamination—retraining language models on their own generated text—affects output diversity across 13 publicly released checkpoints. Using a fixed contamination protocol over five generations, the authors find a wide spread in 4‑gram diversity (0.187 to 0.940), indicating that some models collapse into repetitive fragments while others remain largely unaffected. The study shows that a model’s susceptibility to collapse is an intrinsic property of the checkpoint, not predicted by parameter scale or static indicators, and that simple interventions such as tightening top‑p sampling can significantly slow or halt collapse.
arXiv Computation and Language
Oct 15
From Unreliable to Usable: Decomposing Failure Modes and Confidence Signals in LLM Entity Extraction - Evidence from Cyber Threat Intelligence
arXiv:2610.20417v1 Announce Type: new Abstract: Audits of large language model (LLM) extraction from full-length documents report severe unreliability, leaving two questions open: what the failures a...
arXiv Machine Learning
Jul 9
Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
arXiv:2607. 06596v1 Announce Type: cross Abstract: Trusted monitoring is a central defense in AI control: a cheaper trusted model scores an untrusted model's actions for sabotage, and the most suspicious are audited or deferred.