arXiv:2606. 20959v2 Announce Type: replace Abstract: Language models may encode both outdated facts and their newer replacements.
arXiv:2607. 11889v1 Announce Type: cross Abstract: Large language models trained on unrestricted internet corpora inevitably embed information from the future, introducing lookahead bias that compromises the validity of backtests and causal inference in finance and the social sciences.
arXiv:2607. 11327v1 Announce Type: cross Abstract: Model editing keeps large language models (LLMs) up to date without retraining, but temporal facts expose a limitation of the prevailing locate-and-edit paradigm: an update is not always a replacement.
The paper introduces point‑in‑time language models that are trained only on text available up to each calendar date, thereby eliminating lookahead bias that plagues standard large language models. By scaling decoder‑only transformers to 4 billion parameters and training on 1 trillion chronologically filtered tokens from FineWeb, the authors produce monthly checkpoints from 2013 to 2024 that achieve performance close to unconstrained models of similar size on common‑sense reasoning and language understanding tasks. Instruction fine‑tuning with LoRA further enhances downstream usability, and the authors release the full pipeline—including dataset construction, training infrastructure, and evaluation code—for reproducible research.
arXiv:2608. 08512v1 Announce Type: new Abstract: Evolving documents, such as laws, tax codes, and software documentation, are amended, replaced, and sometimes reverted over time, so a question has different correct answers at different dates.
arXiv:2608.23034v1 Announce Type: cross Abstract: Controlling restricted knowledge in large language models is essential for model alignment and safe deployment. Test-time unlearning avoids costly re...
arXiv:2610.28366v1 Announce Type: new Abstract: Independently trained autoregressive language models often acquire similar capabilities, but it remains unclear whether their internal representations...
The paper introduces Library of Layers (LoL), an architecture that dynamically orders transformer layers during inference, allowing a single model to adapt its computation to the task at hand. LoL’s single hyperparameter—the size of its global library of transformer blocks—enables a smooth transition between parameter reuse and sparsity, unifying dense, depth‑recurrent, and mixture‑of‑experts trade‑offs. Experiments show LoL matches or exceeds these baselines on text, synthetic Sudoku and maze tasks, and demonstrates task‑specific routing in multilingual and safety‑focused settings.
The paper introduces anchored factual recall, where language models must choose an object for a subject–relation pair conditioned on an explicit anchor such as a time. Using causal interventions on four open‑weight models, the authors identify two parallel computational routes: a subject‑mediated route that aligns the subject representation with the queried year, and a direct route that lets the final position attend to the anchor and uses upper‑layer MLPs to bias the answer. These routes combine to determine the model’s answer, showing that factual recall involves selecting among context‑dependent alternatives rather than retrieving a single stored object.
WaveTLM introduces a unified compiler‑executor framework that transforms natural‑language requests into typed task states and reliable numerical or structured outputs for time‑series tasks. The authors present ExecTS‑QA, a benchmark covering forecasting, imputation, classification, anomaly detection, and waveform analysis, and show that a single WaveTLM checkpoint achieves 99.40% contract‑valid coverage versus 37.83% for a string‑first baseline while maintaining balanced predictive performance. Additional evaluations on SciTS, TSQA, IRTS‑ToolBench, and ARFBench demonstrate the model’s transferability.
PRISM is a framework that evaluates how activation interventions transfer across precision, languages, and model scales to mitigate unsafe tool‑calling behaviors in language models. By leveraging lower‑resource settings such as quantized NF4 labels and cross‑lingual transfer, PRISM demonstrates that alignment improvements can be achieved with minimal scenarios and that these gains transfer to held‑out languages and larger models. The study shows that auxiliary supervision from compressed models and multilingual data can effectively uncover and reduce hazardous tool‑calling behaviors across different model families.
arXiv:2609.14168v1 Announce Type: cross Abstract: Context parameterization enables large language models (LLMs) to internalize contexts into reusable model parameters, avoiding repeated processing ac...
arXiv:2609.00184v1 Announce Type: cross Abstract: Large language models (LLMs) rely on static pretraining corpora, causing their knowledge to become outdated over time. Existing approaches for evalua...