Inverse decision modeling infers latent properties of decision processes from observed behavior, but existing formulations rely primarily on action trajectories. In verbalized cognitive tasks, task execution also produces response dynamics that action-only formulations leave unmodeled, such as verbal production, interaction, and hesitation.
arXiv:2608.09222v2 Announce Type: replace Abstract: Inverse decision modeling infers latent properties of decision processes from observed behavior, but existing formulations rely primarily on action...
The paper introduces Process-aware Language Cognitive Diagnosis (PLCD), a framework that replaces traditional ID-based embeddings in Cognitive Diagnosis Models with language-derived structures and response records. PLCD employs large language models to build concept schemas and cognitive process graphs, and uses a Language-to-Cognition Mapper with DA-MoE experts and contrastive learning to map textual evidence into a unified cognitive space. Experiments demonstrate that PLCD outperforms conventional baselines in student performance prediction and shows strong cognitive transfer, improving cold-start robustness and cognitive grounding.
The paper introduces a method that represents language models as log‑likelihood vectors over prompt‑response pairs, enabling the construction of model maps that compare conditional distributions. Squared Euclidean distances in this vector space approximate KL divergence, and experiments show that these maps reveal global structure related to model attributes and task performance. The approach also captures systematic shifts from prompt changes, supports additive compositionality for predicting downstream scores, and offers PMI vectors to mitigate unconditional distribution effects, thereby aiding analysis and prediction of input‑dependent behavior.
arXiv:2607. 29181v1 Announce Type: cross Abstract: Agentic assistants capable of proactive, personalized interactions require structured models of user intent and workflow.
arXiv:2608. 05224v1 Announce Type: new Abstract: Large language models fine-tuned on human behavioural data have emerged as general-purpose cognitive proxies, but the scale this requires, and whether these models process task structure or exploit statistical shortcuts, remain open questions.
The paper introduces CROP, a method for selective on‑policy distillation that prioritizes token‑level supervision based on task relevance. CROP uses paraphrase‑calibrated counterfactual sensitivity to measure how much each response token depends on the semantic content of the input, constructing validated original‑paraphrase‑counterfactual triplets for each prompt. Experiments in two teacher‑student settings show that CROP outperforms other selectors, improving aggregate performance by 1.92 and 2.96 points.
arXiv:2606.03371v4 Announce Type: replace Abstract: Reliable proactive agents must choose an action and judge whether current evidence is sufficient to act. We study retail service from sparse third-...
On-policy distillation (OPD) supervises a student language model on trajectories sampled from its current policy, but assigns equal credit to response tokens with unequal supervision value. Selective OPD addresses this limitation by allocating supervision non-uniformly across response tokens according to their estimated training value.
arXiv:2605. 23055v2 Announce Type: replace-cross Abstract: Frontier language models sometimes recognize that they are being evaluated and adjust their behavior, undermining validity of benchmark results.
Large language models increasingly produce and interpret verbal probability expressions, yet whether these expressions carry consistent meaning across models (or match human perceptions of uncertainty...
Task Vectors (TVs) are compact steering directions for large language models, yet their internal structure has been unclear. This study proposes a decomposition of a TV into Task Recognition (TR), Task Learning (TL), and residual components using candidate-label unembeddings, and optimizes TR and TL coefficients via Bayesian optimization. Across various models, tasks, and TV types, the optimized TR/TL directions outperform original TVs, and isolated interventions on each component yield distinct, functionally consistent logit effects, indicating that TVs can be meaningfully interpreted and controlled.
arXiv:2606. 28186v1 Announce Type: cross Abstract: Predicting human item difficulty is central to educational assessment, where reliable estimates support fairness and effective test construction.