arXiv:2607. 23191v1 Announce Type: new Abstract: Fine-tuned code LLMs can be conditioned on a lightweight design-intent header to steer parametric CAD generation, but whether the model actually reads the header's content has not been tested under a metric independent of the conditioning itself, nor with a causal control.
arXiv:2607. 23191v2 Announce Type: replace Abstract: Fine-tuned code LLMs are often conditioned on a design-intent header to steer parametric CAD generation, but whether the model reads that header's content has been tested neither under execution-level scoring nor with a causal control.
arXiv:2607. 23191v3 Announce Type: replace Abstract: Fine-tuned code LLMs are routinely conditioned on a design-intent specification, but the correctness axis of such a signal -- a wrong intent rather than an absent one -- has not been tested, and the benefit of conditioning is usually scored with the same detector that defines the signal.
The paper investigates why reasoning‑augmented text‑to‑image models like GoT‑R1 sometimes fail on compositional prompts. By separating the explicit textual plan from the decoder, the authors show that the decoder faithfully executes the plan while the planner often writes incorrect spatial relations, especially for phrasing‑dependent cues. Editing or replacing the plan improves image quality without retraining, demonstrating the viability of modular planner‑decoder architectures.
arXiv:2608. 09490v1 Announce Type: new Abstract: Task arithmetic treats fine-tuning displacements as composable directions in weight space, yet it remains unclear when parameter addition reflects predictable changes in model function.
arXiv:2607. 29431v1 Announce Type: new Abstract: Large language models increasingly generate optimization models from natural language, but existing evaluation often reduces a generated model and its ground truth to a single equivalent/not-equivalent verdict or an execution-success rate--labels that are neither independently checkable nor faithful to the multiple distinct senses in which two formulations can agree.
The paper introduces AgentDiff, a metric that quantifies how much LLM agents’ answers differ when inputs are altered by meaning‑bearing rewrites (paraphrases, synonym substitutions) versus presentation changes (reordering, formatting, distractors). Across 68 model–benchmark–scaffold combinations involving ten LLMs and over 1,500 questions, meaning‑bearing rewrites consistently produce a roughly 20‑percentage‑point higher inconsistency rate than presentation changes, a gap that persists across severity proxies and remains significant even outside the Qwen family. Trace analysis reveals that meaning‑bearing rewrites preserve the first action but reduce thought similarity from the second step onward, extending the divergence cascade—a phenomenon termed “stealth divergence.”
The paper introduces the concept of perfect aliasing, where a truth probe that aligns truthful reporting with a task’s prescribed action cannot differentiate between the two based solely on its labels. In a binary reporting game, probes fitted on compliant contexts yield identical optimizations, while on rival contexts their labels are complementary, causing their AUROCs to sum to one across 751 cell-layer pairs. By employing randomized codebooks and mixed-context fitting, the authors demonstrate that separating prescribed output symbols from semantic action enables perfect recovery of truth, achieving an AUROC of 1.000 on rival trials for a reward-trained Gemma-2-9B policy, whereas conventional probes perform near chance.
arXiv:2606. 30449v1 Announce Type: new Abstract: Probes on model internals could help monitor agentic systems if they identify harmful text or tool actions before those actions are generated.
arXiv:2608. 11797v1 Announce Type: new Abstract: Model merging by task arithmetic works until it doesn't, and the field diagnoses why with magnitudes: layerwise representation bias, deviations from cross-task linearity, parameter overlap.
arXiv:2609.07944v1 Announce Type: new Abstract: Existing causal-inference benchmarks for LLMs mostly score method descriptions or whether generated code runs, not whether the executed workflow recove...
arXiv:2608. 02665v1 Announce Type: cross Abstract: A benchmark score is a measurement instrument, yet most benchmarks read each item at a single canonical surface form.
ACE is a self‑correcting agentic canvas editor that operates on a hierarchical scene‑graph rather than flat document formats, enabling reliable multi‑slide presentation automation. It pairs a presentation‑specialized action space of 98 tools with CARE, a content‑aware router that reduces input tokens by about 89%, and a ground‑truth‑free instruction‑following judge that feeds natural‑language critiques back into the agent for self‑correction. In benchmarks, ACE outperforms a comparable agentic HTML pipeline on instruction following (4.23 vs. 3.81), runs 1.75× faster, costs 44% less, and is preferred by 58.7% of blind raters, with 81% favoring the self‑corrected output.