cs.CVNov 12, 2026

Joint Null-Space Projection for Data-Free Model Merging

Authors: Tian Xia

Abstract

Merging independently fine-tuned checkpoints into a single multi-task model requires no data, no labels, and no gradient steps. In practice the task vectors of different tasks interfere, and the standard remedies--averaging, magnitude pruning, sign election--operate entry by entry, ignoring that each delta induces a low-dimensional update on the hidden features it acts upon. We recast merging as a subspace interference problem. For every two-dimensional weight tensor we take the top-r right-singular subspace of each task's delta--the input directions along which that task moves the representation most--and remove from each task vector the component lying in the union of the other tasks' subspaces. The resulting Joint Null-Space Projection (JNSP) is closed-form and order-independent, needs no forward pass, and is defined entirely by the checkpoint set. On eight CLIP image-classification tasks JNSP merges ViT-B/16 experts to 89.45% average accuracy without using a single image, 11.88 points above the strongest parameter-wise training-free baseline (TIES-Merging, 77.57%), 14.22 above Task Arithmetic, and 0.98 above WUDI's data-free consensus merge (88.47%). The gain holds across three backbones on the independently trained TALL release (+10.8, +12.6, +9.1 over TIES for ViT-B/16, ViT-B/32, ViT-L/14) and replicates there at 89.20%; controls confirm the source--a single shared subspace instead of per-task ones loses 2.9 points, and a sequential variant collapses to 69.6%. On RoBERTa-large the same projection underperforms plain task arithmetic, which we report rather than hide. Code is available at https://github.com.

Figures & tables

Appendix figures & tables35 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Label-Free Cross-Task LoRA Merging with Null-Space Compression

    Mar 27, 2026Wonyoung Lee, Wooseong Jeong, Kuk-Jin YoonModel MergingLanguage Model Merging

  2. Merge++: Universal Merge Refinement Through Data-Free Checkpoint Inversion

    Sep 19, 2026Aditya Pola, Vineeth N. BalasubramanianModel Merging

  3. Model Merging by Output-Space Projection

    May 27, 2026Bethan Evans, Benjamin Etheridge, Stephen Roberts +1Quadratic ProgrammingLanguage Model Merging