Where Paths Split: Localized, Calibrated Control of Moral Reasoning in Large Language Models

• URL: http://arxiv.org/abs/2605.03609v1
• Date: Tue, 05 May 2026 10:30:44 GMT
• Title: Where Paths Split: Localized, Calibrated Control of Moral Reasoning in Large Language Models
• Abstract summary: We present Convergent-Divergent Routing, which traces and edits minimal branch points inside transformer blocks.<n>Gating non-target branches at these loci blocks the downstream propagation while leaving upstream computations intact.<n>Experiments on real-life moral dilemmas show that our method reliably achieves preference preference largely preserves general capabilities.
• Score: 15.280774412504101
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models often display heterogeneous moral preferences across settings. We study inference-time steering toward a desired ethical framework while preserving general competence. We present Convergent-Divergent Routing, which traces and edits minimal branch points inside transformer blocks where ethical-framework-related pathways first converge and then diverge. Gating non-target branches at these loci blocks the downstream propagation while leaving upstream computations intact. We find that this intervention alone increases targeted ethical-framework reasoning. To achieve fine-grained control, we adapt Common Spatial Patterns to the residual stream and extract, for each branch-point layer, a pair of directions that discriminate between utilitarian and deontological frameworks. We then introduce Dual Logit Calibration, a closed-form, minimum-$\ell_2$-norm update that moves the residual within this two-dimensional subspace so the resulting directional projections align with user-specified preference weights. Experiments on real-life moral dilemmas show that our method reliably achieves preference calibration and largely preserves general capabilities, outperforming recent baselines while providing an interpretable mechanism.

Related papers

• Dual-Branch Cross-Projection Debiasing through Diffusion-based Disentanglement [66.67463557274358]
  Foundation models trained on biased datasets often rely on spurious correlations between target labels and non-causal attributes.<n>We introduce Confidence-guided Concept Mining (CBCM) to identify reliable spurious attributes without attribute annotations.<n>We propose Dual-branch Cross-projection Debiasing (DCD), a prompt-tuning framework that separates target and spurious representations into two branches.
  arXiv  Detail & Related papers  (2026-06-23T05:28:47Z)
• Does Reasoning Preserve Alignment? On the Trustworthiness of Large Reasoning Models [55.788110316999166]
  We compare reasoning models produced via supervised fine-tuning, RL-based post-training, and distillation against instruction-tuned baselines.<n>We observe that reasoning models often improve on reasoning benchmarks but exhibit alignment regressions.<n>These regressions are consistent with behavioral drift from the instruction-tuned baseline, measured by KL divergence.
  arXiv  Detail & Related papers  (2026-06-09T16:14:27Z)
• Improved techniques for fine-tuning flow models via adjoint matching: a deterministic control pipeline [7.967038299436285]
  We propose a deterministic adjoint matching framework that formulates human preference alignment for flow-based generative models.<n>Building on this perspective, we introduce a truncated adjoint scheme that focuses on the terminal portion of the trajectory.<n>We further generalize the framework beyond standard KL-based regularization, allowing more flexible trade-offs between alignment strength and distributional preservation.
  arXiv  Detail & Related papers  (2026-05-07T17:12:47Z)
• Recall to Predict: Grounding Motion Forecasting in Interpretable Motion Bank [9.873160019979602]
  We propose an end-to-end differentiable framework that grounds predictions in a comprehensive "motion bank"<n>Rather than regressing paths from a blank slate, our architecture dynamically retrieves explicit motion priors using a novel Anchor Retrieval Layer.<n>Our approach eliminates the "black box" of standard latent queries while achieving competitive multi-modal accuracy on the Argoverse 2 and Open Motion datasets.
  arXiv  Detail & Related papers  (2026-05-02T11:31:26Z)
• Defeasible Conditional Obligation in a Two-tiered Preference-based Semantics (Extended Version) [0.0]
  The paper extends a Hansson-Lewis style preference semantics for dyadic deontic logic.<n>A nonmonotonic reasoning mechanism enables previously derived obligations to be withdrawn when new, potentially conflicting information comes in.
  arXiv  Detail & Related papers  (2026-04-28T11:04:26Z)
• Binary Flow Matching: Prediction-Loss Space Alignment for Robust Learning [23.616336786063552]
  Flow matching has emerged as a powerful framework for generative modeling.<n>We identify a latent structural mismatch that arises when it is coupled with velocity-based objectives.<n>We prove that re-aligning the objective to the signal space eliminates the singular weighting.
  arXiv  Detail & Related papers  (2026-02-11T02:02:30Z)
• Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics [81.80010043113445]
  Local weight fine-tuning, LoRA-based adaptation, and activation-based interventions are studied in isolation.<n>We present a unified view that frames these interventions as dynamic weight updates induced by a control signal.<n>Across methods, we observe a consistent trade-off between preference and utility: stronger control increases preference while predictably reducing utility.
  arXiv  Detail & Related papers  (2026-02-02T17:04:36Z)
• Beacon: Single-Turn Diagnosis and Mitigation of Latent Sycophancy in Large Language Models [0.0]
  Large language models internalize a structural trade-off between truthfulness and obsequious flattery.<n>This latent bias, known as sycophancy, manifests as a preference for user agreement over principled reasoning.<n>We introduce Beacon, a single-turn forced-choice benchmark that isolates this bias independent of conversational context.
  arXiv  Detail & Related papers  (2025-10-19T06:36:57Z)
• Ambiguity-aware Point Cloud Segmentation by Adaptive Margin Contrastive Learning [65.94127546086156]
  We propose an adaptive margin contrastive learning method for semantic segmentation on point clouds.<n>We first design AMContrast3D, a method comprising contrastive learning into an ambiguity estimation framework.<n>Inspired by the insight of joint training, we propose AMContrast3D++ integrating with two branches trained in parallel.
  arXiv  Detail & Related papers  (2025-07-09T07:00:32Z)
• Unifying Perplexing Behaviors in Modified BP Attributions through Alignment Perspective [61.5509267439999]
  We present a unified theoretical framework for methods like GBP, RectGrad, LRP, and DTD.<n>We demonstrate that they achieve input alignment by combining the weights of activated neurons.<n>This alignment improves the visualization quality and reduces sensitivity to weight randomization.
  arXiv  Detail & Related papers  (2025-03-14T07:58:26Z)
• Optimal lower bounds for logistic log-likelihoods [1.3124513975412255]
  The logit transform is arguably the most widely-employed link function beyond linear settings. It is still unclear whether tangent lower bounds sharper than quadratic ones can be derived without undermining the tractability of the resulting minorizer. This article addresses such a challenging question through the design and study of a novel piece-wise quadratic lower bound.
  arXiv  Detail & Related papers  (2024-10-14T09:09:33Z)
• Towards Continual Learning Desiderata via HSIC-Bottleneck Orthogonalization and Equiangular Embedding [55.107555305760954]
  We propose a conceptually simple yet effective method that attributes forgetting to layer-wise parameter overwriting and the resulting decision boundary distortion. Our method achieves competitive accuracy performance, even with absolute superiority of zero exemplar buffer and 1.02x the base model.
  arXiv  Detail & Related papers  (2024-01-17T09:01:29Z)
• Taxonomy Adaptive Cross-Domain Adaptation in Medical Imaging via Optimization Trajectory Distillation [73.83178465971552]
  The success of automated medical image analysis depends on large-scale and expert-annotated training sets. Unsupervised domain adaptation (UDA) has been raised as a promising approach to alleviate the burden of labeled data collection. We propose optimization trajectory distillation, a unified approach to address the two technical challenges from a new perspective.
  arXiv  Detail & Related papers  (2023-07-27T08:58:05Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.