The study investigates whether adaptive layer-skipping and repetition programs truly improve language‑model inference or merely exploit input‑blind controls. Using 32 such programs on Qwen3‑4B‑Base and Llama‑3.1‑8B across 4,413 multiple‑choice items, the authors compare real program gains to those from input‑blind perturbations and fixed letter offsets. Results show that input‑blind controls can produce comparable or even larger headroom, and that real programs only outperform controls in specific menu configurations, indicating that the apparent benefits may not be due to the actual layer computation choices.
arXiv:2610.02267v1 Announce Type: new Abstract: Agent harnesses make many small, typed decisions per task: which model to call, which tool to use, whether retrieved text is relevant, whether an input...
arXiv:2606. 16364v1 Announce Type: new Abstract: LLM agents mis-call tools, and the natural guess is that the model failed to see the right tool in a crowded harness.
Typed decision models are built for settings where model outputs are consumed directly by software. Instead of generating free-form text, they return a decision over a predefined set of options. By co...
Prefill‑only decision models evaluate every candidate in a menu in a single forward pass, avoiding decoding and reducing cost by one to two orders of magnitude compared to generative language models. The paper demonstrates that when only the candidate menu changes, the model’s post‑intervention accuracy can be predicted solely from the cached first‑pass distribution using a simple estimator that renormalizes and selects the argmax, without any labels or second pass. Across seven model families, ten datasets, and two task types, this menu‑only intervention prediction is within 4.2 points of actual accuracy, and in one family it is exact, whereas a probability‑level variant fails by 21 points, indicating the property resides in ranking rather than calibrated probabilities.
arXiv:2609.38612v1 Announce Type: new Abstract: As natural language drives more applications, language models increasingly run inside programs as decision components: the program sends them the curre...
The study examines how renaming option labels in typed decision models affects model behavior. By swapping the names of two options (e.g., from 0/1 to no/yes) while keeping the underlying rubrics unchanged, the authors observed a dramatic shift in decision rankings—AUC dropped from .94 to .23 and answer flips increased by 70.4 per hundred. The effect is amplified with more options and depends on the semantic polarity of the labels, yet the models still maintain a zero type‑error rate.
arXiv:2608. 15022v1 Announce Type: new Abstract: Language models hold latent quantities in a form they can report on, and more of a quantity is present in that form when the task requires reusing it flexibly.
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:2606. 25449v1 Announce Type: cross Abstract: A language model's memory can be worse than having no memory at all.
The paper investigates why large‑language‑model coding agents rarely request a second chunk of tool output, focusing on the precision‑at‑1 rate ($p_1$) of the gold item appearing first in the first chunk. In a benchmark of 500 software‑engineering tasks, the authors compare six value functions and find that increasing $p_1$ does not systematically improve downstream accuracy; the agent can recover the correct answer from any position within the chunk. Adding file‑metadata signals to a keyword scorer actually reduces $p_1$, while a parameter‑free keyword scorer improves $p_1$ but still fails to boost overall accuracy.
The study analyzes 373,019 judgments from LLM‑scored benchmarks, decomposing variance into system, item, judge, and interaction components via generalizability theory. It finds that with a single judge, generalizability converges to a ceiling determined by the system‑by‑judge variance, which is substantially lower in pairwise preference settings, allowing one judge to suffice. The research also reveals significant biases in presentation order and highlights that many published win‑rate claims fall below the measured floor of the benchmarks.
arXiv:2607. 17531v1 Announce Type: cross Abstract: Test-time collaboration, including self-consistency, best-of-N selection, critic models, and verifier pipelines, is often credited with broadly improving LLM reasoning, yet its gains are uneven and sometimes negative.