arXiv:2603. 23530v2 Announce Type: replace-cross Abstract: Large language models often fail to satisfy formatting instructions when they must simultaneously perform demanding tasks.
arXiv:2607. 09665v1 Announce Type: new Abstract: Prompt wrappers often differ only in formatting, yet they can change model scores enough to flip leaderboard conclusions.
Practitioners make three prompt-design decisions with almost no controlled evidence behind them: how to format instructions and context (markdown, plain text, prose, or tabular), how many simultaneous instructions a system prompt can carry before compliance degrades, and how much context a model can hold before recall and honesty degrade. We report two controlled experiments crossing all three factors on one held, contamination-free synthetic corpus (the "Book of Veyra," 8,780 uniquely-named entities, deterministically regenerable from a fixed seed), evaluated across five models.
arXiv:2607. 19257v1 Announce Type: cross Abstract: Practitioners make three prompt-design decisions with almost no controlled evidence behind them: how to format instructions and context (markdown, plain text, prose, or tabular), how many simultaneous instructions a system prompt can carry before compliance degrades, and how much context a model can hold before recall and honesty degrade.
arXiv:2606. 00920v1 Announce Type: cross Abstract: Run-level pass rate overstates retry-free coverage by up to 17.
SyntaxBench is a diagnostic benchmark and statistical evaluation framework for character‑level reasoning in large language models, comprising five core tasks—character counting, letter containment, palindrome detection, edit distance, and longest‑string selection—and a harder substring‑extraction stress test called index_to_span. The benchmark uses paired English and random‑string inputs, zero‑, one‑, and four‑shot prompts, and evaluates models from 2B to 32B parameters across multiple reasoning modes. It reports a wide range of metrics, including exact‑match and relaxed accuracy, Cohen’s kappa, McNemar tests, bootstrap confidence intervals, Kendall’s tau, class‑conditional metrics, tokenization analysis, and multiple‑comparison‑corrected tests.
arXiv:2609.07309v1 Announce Type: cross Abstract: Robustness evaluation of large language models (LLMs) remains a critical challenge, particularly in assessing their sensitivity to perturbations in i...
Accuracy changes after language-model self-revision are usually interpreted as changes in reasoning. We show this can fail at the answer-extraction boundary, and test the failure causally rather than only observationally.
arXiv:2608. 12426v1 Announce Type: new Abstract: Large language models are increasingly deployed in settings that require simultaneous adherence to multiple explicit constraints - reasoning structure, safety boundaries, output schemas.
IBBench-Light is a paired evaluation framework that tests language models on both executing procedures and reading text from the same external record. The benchmark uses twelve semantic bases to generate 144 matched pairs per model, with four instruction‑quantized models producing 1,152 greedy responses. Metrics such as Paired Exact‑Contract Accuracy (PECA) reveal that models like Qwen achieve high success on individual prompts but only 97 complete pairs, highlighting the importance of paired evaluation.
The paper introduces a three-level evaluation framework—behavioral deployment, LM-head readout, and probe recoverability—to distinguish whether a language model fails a syntactic test by not encoding structure or by failing to use it. Using a trilingual control-dependency benchmark, the authors find that probe recoverability consistently exceeds LM-head readout, which in turn exceeds behavioral deployment across seven models and three languages, with the largest gap observed in Qwen3-0.6B Instruct. Layer-localized activation patching shows that instruction tuning shifts the decoded layer later, suggesting decoding favors surface shortcuts and that behavioral evaluation understates what models encode while probing alone overstates what they deploy.
Instruction-following ability is critical for deploying large language models in real-world applications, where downstream components depend on the output satisfying specific constraints. Modern deployments increasingly handle the full task in a single LLM call, with one prompt specifying a layered output whose overall artifact, structural sections, and nested fields must each satisfy concrete constraints.