AIIC AI Intelligence Centre

SOURCE-LINKED INTELLIGENCE

OBC-Prune: Outcome-Based Calibration for Large Reasoning Model Pruning

arXiv · AI, language, vision and robotics · article · Sep 15, 2026 · UTC

Large reasoning models (LRMs) generate long chain-of-thought traces before answering, creating significant inference overhead. Pruning can reduce this cost, but its effectiveness depends on the calibration data used to estimate parameter importance. Recent work calibrates on the model's own rollouts instead of generic dataset, but treats all reasoning tokens uniformly, regardless of whether they contribute to successful reasoning. As a result, pruning protects weights by statistical salience rather than by their contribution to correct reasoning, so weights behind erroneous computation survive

Read original source ↗ Open in workspace

recordType
paper
region
Global

Evidence & attribution

First collected: 2026-09-20T08:20:57.646Z. This is not the publication date.