CoolFace
Modelpublic

Montalte/qwen4b-math-nothink-localize

sourceHugging Faceapache-2.0updated 13d agoView on Hugging Face
0likes146downloads
Model Card

Montalte/qwen4b-math-nothink-localize

Unified Qwen3-4B merge artifact for directional math↔code transfer experiments.

FieldValue
BaseQwen/Qwen3-4B-Base@906bfd4b4dc7f14ee4320094d8b41684abff8539
Source specialistmodrill/math-nothink-q4b-20260908
Domainmath
Modenothink
Methodlocalize

Method

localize = Plan B Localize-and-Stitch on source-only MergeBench validation.

  • —sparsity (keep) target: 0.1
  • —lr=1e7, epochs=10, n-shot=64, seed=42, mask-profile=plan_b
  • —task=math
  • —embed / lm_head skipped in mask (body-only stitch; matches prior Plan B protocol)

Uploaded: 2026-09-11.