在联合学习中微调大型语言模型,以公平意识的快速选择
Yalan Jiang1, Zhongliang Li2, Bin Song1
1State Key Laboratory of Integrated Services Networks, Xidian University, Xi'an, 710071, China; The Hangzhou Institute of Technology Xidian University Hangzhou, Hangzhou, China.
概括
大型语言模型 (LLM) 的联合学习得到了FedPSF-LLM的增强,提高了公平性并降低了通信成本. 该框架解决了LLM部署中的隐私和计算挑战.
科学领域:
- 人工智能的人工智能
- 机器学习 机器学习
- 自然语言处理自然语言处理.
背景情况:
- 大型语言模型 (LLM) 需要对部署进行域特定的微调.
- 数据隐私和计算限制是关键的障碍.
- 联合学习 (FL) 允许在私人数据上进行协作调整,保持机密性.
研究的目的:
- 提出FedPSF-LLM,一个新的FL框架,解决FL-LLM中的非IID退化,通信开销和公平性问题.
- 加强在联邦系统中维护隐私和保证公平的LLM部署.
主要方法:
- 提示选择模块 (PSM) 可自适应地选择具有高影响力的提示参数,以降低传输成本.
- 动态权重模块 (DWM) 根据客户贡献和数据差异调整聚合权重.
- 基于注意力的偏差缓解 (ABM) 通过对齐意识重权来纠正聚合偏差.
主要成果:
- FedPSF-LLM提高了公平性,同时在10个NLP任务和4个LLM中保持了强的整体表现.
- 降低了52.1%的精度差异,提高了8.6%的最差客户精度,并缩小了74.4%的客户绩效差距.
- 实现了76.8%的全球准确性,在公平性和沟通效率方面表现优于8个基准值.
结论:
- 在联邦系统中,FedPSF-LLM为保护隐私和保证公平的LLM部署建立了新的范式.
- 该框架有效地缓解了大型语言模型的联合学习中常见的挑战.
相关概念视频
02:07Improving Translational Accuracy
14.1K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
14.1K
02:07Improving Translational Accuracy
3.5K
3.5K
01:14Impression Management Techniques IV: Altercasting
172
Altercasting is a strategic communication technique in which an individual imposes a specific identity or social role onto another person to influence their behavior and shape the interaction. By presuming a role—such as “responsible leader” or “patient person”—altercasting encourages the target to conform to that identity, often aligning their behavior with the expectations associated with the role. The power of this tactic lies in its subtlety; once a role...
172
01:21Frequency-dependent Selection
23.0K
When the fitness of a trait is influenced by how common it is (i.e., its frequency) relative to different traits within a population, this is referred to as frequency-dependent selection. Frequency-dependent selection may occur between species or within a single species. This type of selection can either be positive—with more common phenotypes having higher fitness—or negative, with rarer phenotypes conferring increased fitness.
23.0K
01:12Observational Learning
832
Albert Bandura's observational learning, also known as imitation or modeling, occurs when a person observes and imitates another's behavior. It is a quicker process than operant conditioning. A well-known example is the Bobo doll study, where children who saw an adult acting aggressively towards the doll were more likely to act aggressively when left alone, compared to those who observed a nonaggressive adult. Many psychologists view observational learning as a form of latent learning...
832
01:22Purposive Learning
442
E. C. Tolman emphasized the purposiveness of behavior — the idea that much of our behavior is goal-directed. For instance, employees who aim for a promotion work diligently to meet their targets. Tolman argued that when classical conditioning and operant conditioning occur, the organism acquires certain expectations. In classical conditioning, a child might fear a dog because they expect it to bite. In operant conditioning, a person might consistently work overtime because they expect a...
442