Research map / Federated learning and privacy / Personalized and heterogeneous FL
FLoRG: Federated Fine-tuning with Low-rank Gram Matrices and Procrustes Alignment
Chuiyang Meng, Ming Tang, Vincent W.S. Wong
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 Procrustes alignment after decomposition (ablation) | FLoRG OPT-125M · MRPC | accuracy: 83.14 -> 86.54 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG OPT-125M · QQP | accuracy: 85.85 -> 88.71 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG OPT-125M · MNLI | accuracy: 80.93 -> 87.20 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG OPT-125M · QNLI | accuracy: 86.72 -> 89.69 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG OPT-125M · WNLI | accuracy: 59.81 -> 65.41 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG OPT-125M · RTE | accuracy: 64.32 -> 68.77 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG RoBERTa-large · MRPC | accuracy: 86.50 -> 89.87 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG RoBERTa-large · MNLI | accuracy: 88.93 -> 91.39 | Improved |
| add Procrustes alignment after decomposition (ablation) | FLoRG RoBERTa-large · QNLI | accuracy: 88.62 -> 92.48 | Improved |
| use semi-orthogonal instead of Kaiming initialization (ablation) | FLoRG OPT-125M · MNLI | accuracy: 85.23 -> 87.20 | Improved |
| use semi-orthogonal instead of SVD initialization (ablation) | FLoRG OPT-125M · MNLI | accuracy: 83.19 -> 87.20 | Improved |
| use semi-orthogonal instead of Kaiming initialization (ablation) | FLoRG OPT-125M · RTE | accuracy: 69.31 -> 68.77 | Hurt |
Closest papers by meaning
- Robust Federated Finetuning of LLMs via Alternating Optimization of LoRA (NeurIPS 2025 · similarity 0.92)
- FLoRA: Federated Fine-Tuning Large Language Models with Heterogeneous Low-Rank Adaptations (NeurIPS 2024 · similarity 0.92)
- Ravan: Multi-Head Low-Rank Adaptation for Federated Fine-Tuning (NeurIPS 2025 · similarity 0.92)
- Adaptive LoRA Experts Allocation and Selection for Federated Fine-Tuning (NeurIPS 2025 · similarity 0.91)
- Federated Fine-tuning of Large Language Models under Heterogeneous Tasks and Client Resources (NeurIPS 2024 · similarity 0.91)
- Selective Aggregation for Low-Rank Adaptation in Federated Learning (ICLR 2025 · 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.
