# LOCKR targets 'stable-but-wrong lock-in' in diffusion language models

> An arXiv preprint proposes LOCKR, a hidden-state trajectory planner that detects and repairs stable-but-wrong answers in diffusion language models, claiming 2.21–5.37 point accuracy gains.

- **Topic**: Models
- **Published**: 2026-09-24T07:49:13.194Z
- **Canonical URL**: https://highsignal.sh/stories/lockr-targets-stable-but-wrong-lock-in-in-diffusion-language-models-27e19e96

## Why It Matters

Diffusion language models commit to answers while denoising is still underway, and the paper claims the usual decoding signals can't tell a confident wrong answer from a right one — pointing to hidden trajectories as a repair signal that may generalize to other test-time reasoning fixes.

## Key Findings & Analysis

### The paper

A new arXiv preprint (2609.27220) from ArXiv CS.CL describes LOCKR, a hidden-state trajectory-guided planner for diffusion language models. The authors identify a reasoning failure they call stable-but-wrong lock-in: an answer stabilizes early around an incorrect value while substantial denoising remains. They report that surface-level decoding signals — confidence, entropy, margin, and answer stability — are not sufficient to reliably separate correct from erroneous lock-in. LOCKR frames selective reasoning repair as a lightweight test-time planning problem: it decides when to spend extra computation, expands targeted repair branches, and picks a continuation using trajectory-aware verification.

### Reported results

Across two diffusion language models and three mathematical reasoning benchmarks, the authors report that hidden-state trajectories consistently beat surface signals and single hidden snapshots at both detecting wrong lock-in and selecting repairs. On natural evaluation distributions, the paper claims absolute accuracy gains of 2.21 to 5.37 percentage points across all five evaluated settings, with repair rates between 22% and 41%. These are the authors' own results on a preprint, not independently verified; the benchmarks are mathematical reasoning only.

## Primary Sources & Citations

- [LOCKR: A Hidden-State Trajectory-Guided Planner for Detecting and Repairing Stable-but-Wrong Lock-In in Diffusion Language Models](https://arxiv.org/abs/2609.27220) — *ArXiv CS.CL (Computation and Language)* (Reporting)

---

[← Back to front page](https://highsignal.sh/) | [Daily Brief](https://highsignal.sh/brief) | [All stories](https://highsignal.sh/latest)