Research map / Architectures and efficiency / Pruning and efficient tuning
DC-Merge: Improving Model Merging with Directional Consistency
Han-Chen Zhang, Zi-Hao Zhou, Mao-Lin Luo, Shimin Di, Min-Ling Zhang, Tong Wei
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add energy smoothing (averaging) (ablation) | DC-Merge (LoRA, ViT-B-32) · 8 tasks | average normalized accuracy: 69.27 -> 73.09 | Improved |
| use linear smoothing instead of averaging (ablation) | DC-Merge (LoRA, ViT-B-32) · 8 tasks | average normalized accuracy: 73.09 -> 73.90 | Improved |
| add energy smoothing (linear) over no smoothing (ablation) | DC-Merge (LoRA, ViT-B-32) · 16 tasks | average normalized accuracy: 74.47 -> 75.80 | Improved |
| add structural mask (ablation) | DC-Merge (FFT, ViT-B-32) · 20 tasks | average normalized accuracy: 77.63 -> 88.18 | Improved |
| add structural mask (ablation) | DC-Merge (FFT, ViT-B-32) · 8 tasks | average normalized accuracy: 87.98 -> 93.55 | Improved |
| add structural mask (ablation) | DC-Merge (LoRA, ViT-B-32) · 8 tasks | average normalized accuracy: 73.61 -> 73.90 | Improved |
| add structural mask (ablation) | DC-Merge (LoRA, ViT-B-32) · 16 tasks | average normalized accuracy: 74.42 -> 75.80 | Improved |
| add energy smoothing to Task Arithmetic (ablation) | Task Arithmetic (ViT-B-32, LoRA) · 8 tasks | average normalized accuracy: 61.73 -> 69.12 | Improved |
| add cover space merging to Task Arithmetic (ablation) | Task Arithmetic (ViT-B-32, LoRA) · 8 tasks | average normalized accuracy: 61.73 -> 68.13 | Improved |
| add energy smoothing on top of cover space merging (ablation) | Task Arithmetic (ViT-B-32, LoRA) · 8 tasks | average normalized accuracy: 68.13 -> 73.82 | Improved |
| add cover space merging to TIES (ablation) | TIES (ViT-B-32, LoRA) · 16 tasks | average normalized accuracy: 70.06 -> 74.47 | Improved |
| add energy smoothing on top of cover space merging (ablation) | TIES (ViT-B-32, LoRA) · 16 tasks | average normalized accuracy: 74.47 -> 75.80 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- RobustMerge — RobustMerge: Parameter-Efficient Model Merging for MLLMs with Direction Robustness (NeurIPS 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- No Task Left Behind: Isotropic Model Merging with Common and Task-Specific Subspaces (ICML 2025 · similarity 0.92)
- Merging on the Fly Without Retraining: A Sequential Approach to Scalable Continual Model Merging (NeurIPS 2025 · similarity 0.91)
- Whoever Started the interference Should End It: Guiding Data-Free Model Merging via Task Vectors (ICML 2025 · similarity 0.91)
- Towards Minimizing Feature Drift in Model Merging: Layer-wise Task Vector Fusion for Adaptive Knowledge Integration (NeurIPS 2025 · similarity 0.91)
- Twin-Merging: Dynamic Integration of Modular Expertise in Model Merging (NeurIPS 2024 · similarity 0.90)
- Dynamic Model Merging Made Slim (NeurIPS 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
