Research map / Architectures and efficiency / Pruning and efficient tuning
MiSS: Revisiting the Trade-off in LoRA with an Efficient Shard-Sharing Structure
Jiale Kang, Qingyu Yin
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 |
|---|---|---|---|
| raise MiSS rank from 16 to 128 (ablation) | MiSS on Llama2-7B · GSM8K | accuracy: 45.90 -> 53.49 | Improved |
| raise MiSS rank from 16 to 128 (ablation) | MiSS on Llama2-7B · Math | accuracy: 3.77 -> 10.08 | Improved |
| replace LoRA with MiSS | RoBERTa-base · GLUE CoLA | score: 62.40 -> 72.86 | Improved |
| replace PiSSA with MiSS | RoBERTa-base · GLUE CoLA | score: 67.28 -> 72.86 | Improved |
| replace PiSSA with MiSS | RoBERTa-base · GLUE MNLI | accuracy: 85.72 -> 85.71 | No clear effect |
| replace LoRA with MiSS | Llama2-7B · NLG average | average score: 24.72 -> 29.30 | Improved |
| replace LoRA with MiSS | Qwen3-4B · Math | accuracy: 15.20 -> 34.82 | Improved |
| replace PiSSA with MiSS | Mistral-7B · NLG average | average score: 44.45 -> 47.79 | Improved |
| replace PiSSA with MiSS | Llama2-13B · NLG average | average score: 39.52 -> 42.11 | Improved |
| replace PiSSA with MiSS | Qwen3-4B · GSM8K | accuracy: 85.78 -> 85.52 | Hurt |
| replace LoRA with MiSS | ViT-Base on VTAB-1K · image average | accuracy: 88.08 -> 88.02 | No clear effect |
| replace LoRA with MiSS | Llama-3.2-3B · PEFT benchmark | test accuracy: 0.4822 -> 0.5080 | Improved |
Later papers that compared with this method
1 later accepted paper reports MiSS as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- SliceFine: The Universal Winning-Slice Hypothesis for Pretrained Networks (ICML 2026 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- PiSSA — PiSSA: Principal Singular Values and Singular Vectors Adaptation of Large Language Models (NeurIPS 2024) · better on 8 of 8 shared results
- DoRA — DoRA: Weight-Decomposed Low-Rank Adaptation (ICML 2024) · better on 6 of 6 shared results
- VeRA — VeRA: Vector-based Random Matrix Adaptation (ICLR 2024) · better on 1 of 1 shared result
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
- LoRA-GA: Low-Rank Adaptation with Gradient Approximation (NeurIPS 2024 · similarity 0.92)
- Flat-LoRA: Low-Rank Adaptation over a Flat Loss Landscape (ICML 2025 · similarity 0.92)
- LoRA-Pro: Are Low-Rank Adapters Properly Optimized? (ICLR 2025 · similarity 0.91)
- Uni-LoRA: One Vector is All You Need (NeurIPS 2025 · similarity 0.91)
- AuroRA: Breaking Low-Rank Bottleneck of LoRA with Nonlinear Mapping (NeurIPS 2025 · similarity 0.91)
- GraLoRA: Granular Low-Rank Adaptation for Parameter-Efficient Fine-Tuning (NeurIPS 2025 · similarity 0.91)
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.
