arXiv:2608. 16643v1 Announce Type: cross Abstract: Automated detection of errors in clinical documentation is a promising application of large language models (LLMs), yet decisions to deploy such models rest on benchmarks that evaluate each clinical note in isolation.
arXiv:2511. 00382v2 Announce Type: replace-cross Abstract: Organizations increasingly adapt Large Language Models (LLMs) from public repositories such as HuggingFace to downstream tasks.
arXiv:2609.37858v1 Announce Type: new Abstract: Many localized large language model (LLM) unlearning methods select a small parameter subset from a localization signal and keep it fixed during optimi...
arXiv:2607. 26654v1 Announce Type: cross Abstract: Post-training alignment is often shallow, eroding under fine-tuning.
arXiv:2607. 01239v1 Announce Type: cross Abstract: Character-level perturbations bypass safety alignment in modern LLMs despite leaving prompts human-readable.
arXiv:2606. 31630v1 Announce Type: new Abstract: Language models increasingly write probabilistic programs (in NumPyro, Stan, or Pyro), but a program that compiles, runs, and passes every unit test can still be \emph{statistically} wrong -- a Gaussian likelihood for heavy-tailed data, a Poisson for over-dispersed counts, an invalid prior support, or a pathological parameterization.
Neurosymbolic Alignment couples a 7B clinical language model with a graph‑based physiological world model to score candidate responses using homeostatic constraints, multi‑hop plausibility, and drug‑interaction penalties. This training‑time framework drives iterative on‑policy updates and achieves a 90.8% Clinical Safety Score on the CSB benchmark, outperforming ORPO, GPT‑4 (5‑shot), and a self‑correction pipeline. Ablation studies show that the HGNN scoring and iterative training are the key contributors to the safety gains.
arXiv:2606. 09165v1 Announce Type: new Abstract: Safety judges are increasingly deployed to evaluate model outputs against evolving criteria, yet recent meta-evaluation work shows they remain brittle under prompt and rubric variation, with false negative-rate swings of up to 0.
The paper introduces SynthSentry, a model‑agnostic method for detecting synthetic data contamination in language‑model training corpora. It computes a distributional divergence score based on lexical diversity collapse, n‑gram tail truncation, and perplexity variance across reference models, requiring no access to the generating model or synthetic labels. Experiments on English corpora contaminated by small open‑weight generators and an instruction‑tuned model show that SynthSentry ranks contamination severity accurately, maintains low false‑positive rates after calibration, and does not degrade downstream fine‑tuning performance at the tested scale.
Warning: This paper studies stereotypes and biases, and contains potentially disturbing examples, used for illustration purposes only. Our findings should not be interpreted as an argument against alignment.
Automated detection of errors in clinical documentation is a promising application of large language models (LLMs), yet decisions to deploy such models rest on benchmarks that evaluate each clinical note in isolation. Error-detection benchmarks are typically constructed by injecting errors into notes, such that each erroneous note has a natural counterpart.
Manacá-1B is a 1.72‑billion‑parameter, open decoder‑only language model trained from scratch for Brazilian Portuguese, released with a fully containerized, reproducible training pipeline and complete logs. The authors evaluate it against nine open baselines on four Portuguese benchmarks, reporting standard errors and paired significance tests, and find that Manacá-1B outperforms smaller models on LAMBADA‑PT while remaining competitive on commonsense completion. They also uncover a tokenizer‑related evaluation pitfall that can drastically lower accuracy and provide a simple fix, releasing all code, logs, and corrected tokenizer for full reproducibility.