{"id":"arxiv/2611.21432","url":"https://arxiv.org/abs/2611.21432","tags":["cs.LG","cs.CL","math.OC"],"score":0,"title":"Just sort it! A note on the LP behind rank-one anti-loop edits","domain":"cs.LG","source":"arxiv","authors":["Matteo Fischetti"],"reviews":0,"subject":"cs.LG","summary":"Weight-space programming removes the thinking loops of a small reasoning model by a rank-one edit W' = W + u v^T of an MLP output matrix, where u is a loop direction and v is the solution of a sparse Linear Program (LP) over the neuron activations. We investigate whether the LP is needed. We replace it by a sort: we take the 8 neurons whose activation has the smallest coefficient of variation, and give them equal shares of the push that the LP solution realizes. On Qwen3.5-0.8B the sorted edit scores 67.0% on 1000 problems (MATH-500 and GSM8K) against 67.4% for the LP edit and 9.5% for the…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-11-16","reactions":{}}