Xungang Gu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xungang Gu has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Chao Chen
Shandong Artificial Intelligence Institute, Qilu University of Technology (Shandong Academy of Sciences), Jinan 250014, China
1 shared publication Chong Di
1 shared publication- Chong Di
Shandong Artificial Intelligence Institute, Qilu University of Technology (Shandong Academy of Sciences), Jinan, China
1 shared publication - He Zhang
Kexin Technology
1 shared publication Jiandong Lv
1 shared publication- Jiaze Sun
Kexin Technology, Beijing 100012, China
1 shared publication
Latest Publications
- Open Access
- 1 December 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 114
Multi-stage temporal and cross-view contrastive learning for self-supervised multi-lead ECG classification
- 1 December 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 114
Open AccessMulti-stage temporal and cross-view contrastive learning for self-supervised multi-lead ECG classification
- 22 September 2025
- Published by Elsevier in Measurement
- Vol. 257
Thrust force synthesis-based motion control for a voice-coil actuator on a reciprocating rig with lumped disturbing-force compensation
- 22 September 2025
- Published by Elsevier in Measurement
- Vol. 257
Thrust force synthesis-based motion control for a voice-coil actuator on a reciprocating rig with lumped disturbing-force compensation
- Open Access
A Comprehensive Approach to Instruction Tuning for Qwen2.5: Data Selection, Domain Interaction, and Training Protocols
Instruction tuning plays a pivotal role in aligning large language models with diverse tasks, yet its effectiveness hinges on the interplay of data quality, domain composition, and training strategies. This study moves beyond qualitative assessment to systematically quantify these factors through extensive experiments on data selection, data mixture, and training protocols. By quantifying performance trade-offs, we demonstrate that the implicit method SuperFiltering achieves an optimal balance, whereas explicit filters can induce capability conflicts. A fine-grained analysis of cross-domain interactions quantifies a near-linear competition between code and math, while showing that tool use data exhibits minimal interference. To mitigate these measured conflicts, we compare multi-task, sequential, and multi-stage training strategies, revealing that multi-stage training significantly reduces Conflict Rates while preserving domain expertise. Our findings culminate in a unified framework for optimizing instruction tuning, offering actionable, data-driven guidelines for balancing multi-domain performance and enhancing model generalization, thus advancing the field by providing a methodology to move from intuition to systematic optimization.
Open AccessA Comprehensive Approach to Instruction Tuning for Qwen2.5: Data Selection, Domain Interaction, and Training Protocols
Instruction tuning plays a pivotal role in aligning large language models with diverse tasks, yet its effectiveness hinges on the interplay of data quality, domain composition, and training strategies. This study moves beyond qualitative assessment to systematically quantify these factors through extensive experiments on data selection, data mixture, and training protocols. By quantifying performance trade-offs, we demonstrate that the implicit method SuperFiltering achieves an optimal balance, whereas explicit filters can induce capability conflicts. A fine-grained analysis of cross-domain interactions quantifies a near-linear competition between code and math, while showing that tool use data exhibits minimal interference. To mitigate these measured conflicts, we compare multi-task, sequential, and multi-stage training strategies, revealing that multi-stage training significantly reduces Conflict Rates while preserving domain expertise. Our findings culminate in a unified framework for optimizing instruction tuning, offering actionable, data-driven guidelines for balancing multi-domain performance and enhancing model generalization, thus advancing the field by providing a methodology to move from intuition to systematic optimization.