HEXA: Heterogeneity-aware Exact Aggregation for Efficient Fine-Tuning in Federated Learning
Bibliographic record
Abstract
Federated Learning (FL) combined with Parameter-Efficient Fine-Tuning (PEFT) methods, such as Low-Rank Adaptation (LoRA) has emerged as a promising approach to address data scarcity challenges in fine-tuning Large Language Models (LLMs) while ensuring privacy and computational efficiency. However, when applying LoRA in traditional FL, separately averaging the adapters during aggregation results in non-exact aggregation. While recent research has investigated this issue, its application to heterogeneous data settings remains largely unexplored. Data heterogeneity across clients can significantly affect the effectiveness of parameter-efficient adaptations and complicate the aggregation process.In this work, we explore the concept of exact aggregation in heterogeneous federated fine-tune settings, specifically focusing on LoRA-based approaches. We propose HEXA (Heterogeneity-aware EXact Aggregation), a novel method that mitigates the effects of data heterogeneity while preserving the benefits of exact aggregation in LoRA-enabled FL. We present a comprehensive theoretical framework for extending exact aggregation to heterogeneous settings and validate our approach through extensive empirical evaluation on the GLUE benchmark. Our results show that HEXA improves model performance in heterogeneous contexts while maintaining the computational efficiency of PEFT methods.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.018 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.003 | 0.004 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".