arxiv arXiv cs.CL · 9d ago

Post-Hoc Operators Fail to Improve Accuracy in Small Code Models

A measurement study finds that 26 semantic post-hoc operators do not improve held-out accuracy over Best-of-N in frozen small code models. While two operators—expression-layer recovery and adaptive consensus early-stop—offer benefits in compute efficiency or program recovery, none outperform BoN in accuracy. The results highlight systemic limitations in error detection and coverage, suggesting that model harnesses and error coverage must be improved before post-hoc reasoning is considered.

arxiv arXiv cs.CL · 9d ago

TokenPilot: Cache-Efficient Context Management for LLM Agents

TokenPilot reduces inference costs by 61% to 87% in both isolated and continuous modes, outperforming prior systems in cost efficiency while maintaining competitive performance. It uses ingestion-aware compaction and lifecycle-aware eviction to preserve prompt cache continuity and minimize token footprints.

arxiv arXiv cs.CL · 9d ago

MetaSyn: Benchmarking LLM Agents on Meta-Analysis Articles

MetaSyn introduces a dataset of 442 expert-curated meta-analyses from Nature Portfolio. It evaluates twelve LLM agent configurations and reveals a critical bottleneck in study screening, where no system recovers more than 52.7% of ground-truth included literature despite high retrieval recall.

arxiv arXiv cs.CL · 9d ago

Language Models Encode Value of Their Current Trajectory

Qwen3-8B internally tracks the value of its current trajectory, defined as the likelihood of achieving its goals. This 'value' axis distinguishes confidence levels, backtracking behavior, and code correctness, and shows that preference optimization boosts confidence in rewarded behaviors. The model assigns low value to politically sensitive queries post-training, and fine-tuning increases confidence within specific domains.

arxiv arXiv cs.AI · 9d ago

Semantic Flip: Synthetic OOD Generation for Robust Refusal

Semantic Flip proposes a framework to synthesize out-of-distribution samples by transforming queries and video memory to create unanswerable pairs. These pairs train a lightweight rejection module that attaches to existing vision-language models without retraining, improving refusal performance in embodied question answering and spatial localization. On the new SpaceReject benchmark, it achieves an F1 score of 0.9559.

arxiv arXiv cs.AI · 9d ago

Variance in LLM Circuit Discovery: Causes and Mitigations

This paper analyzes variance in circuit discovery for large language models, identifying resampling, rephrasing, and sample-wise variance. It shows CEAP reduces resampling variance and argues rephrasing variance stems from prompt templates activating different circuits, implying LLMs may be inherently hard to steer. The study also finds sparsity does not resolve these issues and that sample-wise variance is largely benign due to selective contribution scaling affecting unfaithfulness scores.

arxiv arXiv cs.AI · 9d ago

MA-SBI: Calibration-Free SBI via Side-Channel Guidance

MA-SBI introduces a calibration-free simulation-based inference framework that uses side-channel text, like regime labels or instructions, to correct for simulator misspecification. It employs a learned corrector to apply observation-space shifts before posterior inference, without needing ground-truth parameter pairs or retraining. On hide-the-calibration benchmarks, MA-SBI matches the oracle posterior with text alone, outperforming RoPE under limited data, and shows robustness on real-world epidemiological and cognitive-science datasets.

arxiv arXiv cs.AI · 9d ago

RAID: Semantic Graph Diffusion for True Cold-Start and Cross-Lingual Forecasting

RAID introduces a framework that uses metadata-driven semantic retrieval and graph-conditioned diffusion to address true cold-start scenarios. It outperforms foundation models and baselines in forecasting accuracy and interval coverage, reduces inference latency significantly, and enables zero-shot cross-lingual transfer via a shared semantic space.

arxiv arXiv cs.AI · 9d ago

Unified Causal-Origin Taxonomy for Distributional Shifts in RL

This paper introduces a unified causal-origin taxonomy that categorizes distributional shifts in reinforcement learning into internal, agent-driven, and external, environment-driven sources. It unifies ID/OOD generalization and non-stationary settings by framing shifts as structured changes in the agent-environment interaction process, using a POMDP decomposition and a shifted-time boundary perspective.

arxiv arXiv cs.AI · 9d ago

CircuitLasso: Scalable Circuit Learning for LLM Interpretability

CircuitLasso proposes a scalable method for learning sparse circuits in large language models using sparse linear regression. It achieves structural accuracy comparable to state-of-the-art intervention-based methods at significantly lower computational cost, while enabling efficient discovery of semantic feature propagation and improving performance on domain-generalization tasks with reduced cost.

arxiv arXiv cs.AI · 9d ago

Causal Model of Theory of Mind in AI Conflict

This paper proposes a structural causal model using a directed acyclic graph to define when Theory of Mind engagement is causally warranted in human-machine conflict. The model identifies four exogenous conditions, five mediators, and three causal pathways for ToM activation, with epistemic accuracy as the primary outcome. It offers a resource-rational framework for AI social reasoning, validated through simulation and human-machine studies.

arxiv arXiv cs.AI · 9d ago

Causal Framework for Auditing Synthetic Data Disclosures

A new empirical auditing framework detects and classifies synthetic data disclosures as either true or phantom. It distinguishes direct reproductions of user data from incidental generation without model access or training, using only synthetic output and a held-out control set. The method provides tighter privacy leakage bounds than prior approaches and requires significantly fewer computational resources.

arxiv arXiv cs.AI · 9d ago

Low Frame Rate Degradation in Neural Audio Codecs

A quality cliff at 6.25 Hz in neural audio codecs is caused by insufficient training token exposure due to fixed clip duration. Correcting this training configuration enables smooth WER degradation down to 3.1 Hz and 1.6 Hz, indicating low frame rate efficiency is more achievable than previously thought.

arxiv arXiv cs.AI · 9d ago

Textual Reviews Have Limited Impact in Recommendation Models

A study finds that while textual review signals can be fused with collaborative data, their marginal contribution remains limited compared to collaborative signals in matrix factorization models. Adaptive fusion and cross-attention mechanisms improve representation flexibility, but do not significantly boost performance across datasets.

arxiv arXiv cs.AI · 9d ago

AI research documentation improves over decade

Analysis of 56,800 AI conference papers shows documentation practices improved from 2014 to 2024. Papers sharing both code and data increased from 11% to 64%, and estimated reproducibility rose from 28% to 64%. These improvements predate formal reproducibility checklists, indicating a broader shift toward open science.

arxiv arXiv cs.AI · 9d ago

Agentic LLM Framework for HTS Code Classification

A consensus-based agentic large language model framework is proposed for accurate 10-digit Harmonized Tariff Schedule code classification in Canadian maritime logistics. Evaluated on 3,300 expert-labeled product records, the framework shows that fine-grained HTS classification remains challenging for advanced LLMs, highlighting the need for evidence-grounded, uncertainty-aware, and human-in-the-loop workflows.

arxiv arXiv cs.AI · 9d ago

ActiveSAM: Fast and Accurate Open-Vocabulary Segmentation

ActiveSAM is a training-free, zero-shot framework that enhances SAM 3 for open-vocabulary semantic segmentation by identifying an image-conditioned active class set. It improves speed-accuracy tradeoff, outperforming SegEarth-OV3 by +1.4 mIoU on average and running up to 5.5x faster on large-vocabulary datasets, with strong robustness under image corruption.

arxiv arXiv cs.AI · 9d ago

Bayesian Audits Reveal Inconsistent AI Evaluation Timelines

Public AI evaluation archives show that a single terminal result can arise from two distinct pre-terminal histories, with estimated times to reach 95% of performance ceilings at 23.03 or 75.13. A candidate selection-aware frontier model fails synthetic recovery and uncertainty calibration, and is rejected by fixed audit gates. An archive-and-adjudication protocol verifies timing boundaries and falsifies unsupported frontier claims.

arxiv arXiv cs.AI · 9d ago

TuneJury: Open Metric for Music Generation Preference Alignment

TuneJury is an open, instance-level pairwise reward model that predicts music preference scores from text prompts and audio clips. It is trained on diverse human-preference data and demonstrates strong generalization, with anchor calibration enabling efficient post-hoc alignment for music generation systems.

arxiv arXiv cs.AI · 9d ago

TokenPilot: Cache-Efficient Context Management for LLM Agents

TokenPilot reduces inference costs by 61% to 87% in both isolated and continuous modes, outperforming prior systems in cost efficiency while maintaining competitive performance. It uses ingestion-aware compaction and lifecycle-aware eviction to preserve prompt cache continuity and minimize token footprint without introducing prefix mismatches.