Anatomy of a Sound Neural Reasoner: One-Shot Amortization, First-Pass Poisoning, and Search Inertness in Clue-Rich Completion

• URL: http://arxiv.org/abs/2607.19635v1
• Date: Wed, 22 Jul 2026 00:05:34 GMT
• Title: Anatomy of a Sound Neural Reasoner: One-Shot Amortization, First-Pass Poisoning, and Search Inertness in Clue-Rich Completion
• Abstract summary: In clue-rich Sudoku, one forward pass commits essentially the entire grid.<n>All hard-slice failures are decided before search begins, when the first pass confidently deletes a value required by the true solution.<n>In clue-rich completion, LDT-like systems are one-shot amortized predictors rather than learned search procedures.
• Score: 0.0
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Neural solvers are built to deduce, branch, and revise intermediate states. The Lattice Deduction Transformer (LDT) appears to do exactly that. In clue-rich Sudoku, it does not: one forward pass commits essentially the entire grid (every blank cell on standard 6x6, 94-96% on augmented 9x9), turning the iterative solver into a one-shot predictor wrapped in an exact verifier. All hard-slice failures are decided before search begins, when the first pass confidently deletes a value required by the true solution. We call this first-pass poisoning. Adding learned branching, MRV, backtracking, value exclusion, and shared nogoods (CoLT) does not change which Sudoku instances are solved; it cuts repeated invalid derivations 1,497-fold. At the frozen training budget, constraint-graph attention alone matches full-CoLT accuracy, while positional tables recover only under substantially longer training, indicating an optimization and sample-efficiency advantage rather than an absolute capacity difference. The diagnosis predicts two effective interventions. Digit-permutation augmentation raises 9x9 accuracy from below 1% to 96.5 +/- 0.3 across three training seeds on a symmetry-disjoint split. Test-time union over symmetry-transformed passes raises all three hard-slice checkpoints from 72.8-78.9% to 100% without retraining. On from-scratch graph coloring, one-shot behavior disappears and search changes accuracy. In clue-rich completion, LDT-like systems are one-shot amortized predictors rather than learned search procedures: accuracy is determined by calibration and symmetry, while search primarily removes computational waste.

Related papers

• Lagged Coupling: Internal Representations Become Readable Before They Become Causal [0.0]
  Internal readability systematically outruns causal efficacy, and the lag does not shrink with scale.<n>A pre-registered OLMo-2 replication preserves the direction at attenuated magnitude.
  arXiv  Detail & Related papers  (2026-09-01T10:46:42Z)
• The Equilibrium Is the Initialization: Lazy Identity Collapse in Physics-Structured Deep Equilibrium Reasoning [0.0]
  Deep equilibrium models promise input-adaptive implicit computation.<n>We report a cautionary study of a port-Hamiltonian DEQ with learned reasoning tasks.<n>We show that the standard zeroing ablation is confounded and gives wildly seed-dependent answers.
  arXiv  Detail & Related papers  (2026-07-13T05:45:35Z)
• When Agents Commit Too Soon: Diagnosing Premature Commitment in LLM Agents [0.0]
  Long-answer LLM agents can fail quietly: they settle on one reading of the evidence early, then spend the rest of the run defending it.<n>We define representational commitment as cross-run hidden-state convergence at a fixed reasoning step.<n>A runtime monitor detects inconsistent trajectories from hidden states at AUROC up to 0.97 (0.85--0.88 under a stricter split)
  arXiv  Detail & Related papers  (2026-06-22T07:13:13Z)
• A Verifiable Search Is Not a Learnable Chain-of-Thought [0.0]
  It is tempting to assume any task solvable by a short program can be taught to a model as its chain-of-thought.<n>This paper shows the assumption fails for an identifiable class of procedures.
  arXiv  Detail & Related papers  (2026-06-20T04:52:14Z)
• Right Makes Might: Aligning Verified Hidden States Empowers RL Reasoning [55.264863369127774]
  Current methods reduce each correct rollout to a single reward bit, ignoring the geometric structure shared among their hidden states.<n>We propose Hidden-Align, an auxiliary loss function that aligns the last-layer hidden states of correct rollouts at the anchor token during RL training, with zero overhead in both training and inference.<n>On eight mathematical reasoning benchmarks, Hidden-Align improves average pass@1 over the DAPO baseline by 3.8, 6.2, and 5.4 percentage points on Qwen3-1.7B, 4B, and 14B respectively, with consistent pass@k gains across all three scales.
  arXiv  Detail & Related papers  (2026-06-02T06:51:15Z)
• Rethinking Dense Sequential Chains: Reasoning Language Models Can Extract Answers from Sparse, Order-Shuffling Chain-of-Thoughts [51.84894623128418]
  Modern reasoning language models generate dense, sequential chain-of-thought traces implicitly assuming that every token contributes and that steps must be consumed in order.<n>We challenge both assumptions through a systematic intervention pipeline--removal, masking, shuffling, and noise injection--applied to model-generated reasoning chains.<n>We find that answer extraction operates on a sparse, order-insensitive, and structurally robust informational substrate.
  arXiv  Detail & Related papers  (2026-05-08T06:15:50Z)
• Correction and Corruption: A Two-Rate View of Error Flow in LLM Protocols [51.56484100374058]
  We propose a paired-outcome measurement interface for auditing a single protocol step on exact-match tasks.<n>For each instance, the interface records a baseline correctness bit and a post-step correctness bit.<n>These rates predict accuracy changes and define a reusable empirical interface testable across seeds, mixtures, and pipelines.
  arXiv  Detail & Related papers  (2026-04-20T13:25:40Z)
• Beta-Scheduling: Momentum from Critical Damping as a Diagnostic and Correction Tool for Neural Network Training [0.0]
  Standard neural network training uses constant momentum (typically 0.9), a convention dating to 1964 with limited theoretical justification for its optimality.<n>This beta-schedule requires zero free parameters beyond the existing learning rate schedule.<n>On ResNet-18/CIFAR-10, beta-scheduling delivers 1.9x faster convergence to 90% accuracy compared to constant momentum.
  arXiv  Detail & Related papers  (2026-03-30T18:53:03Z)
• BadCLIP++: Stealthy and Persistent Backdoors in Multimodal Contrastive Learning [73.46118996284888]
  Research on backdoor attacks against multimodal contrastive learning models faces two key challenges: stealthiness and persistence.<n>We propose BadCLIP++, a unified framework that tackles both challenges.<n>For stealthiness, we introduce a semantic-fusion QR micro-trigger that embeds imperceptible patterns near task-relevant regions.<n>For persistence, we stabilize trigger embeddings via radius shrinkage and centroid alignment.
  arXiv  Detail & Related papers  (2026-02-19T08:31:16Z)
• Predicting Overtakes in Trucks Using CAN Data [51.28632782308621]
  We investigate the detection of truck overtakes from CAN data. Our analysis covers up to 10 seconds before the overtaking event. We observe that the prediction scores of the overtake class tend to increase as we approach the overtake trigger.
  arXiv  Detail & Related papers  (2024-04-08T17:58:22Z)
• Improve Agents without Retraining: Parallel Tree Search with Off-Policy Correction [63.595545216327245]
  We tackle two major challenges with Tree Search (TS) We first discover and analyze a counter-intuitive phenomenon: action selection through TS and a pre-trained value function often leads to lower performance compared to the original pre-trained agent. We introduce Batch-BFS: a GPU breadth-first search that advances all nodes in each depth of the tree simultaneously.
  arXiv  Detail & Related papers  (2021-07-04T19:32:24Z)

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.