Abstract:The increasing adoption of web crawling opt-outs by copyright holders of online content raises critical questions about the impact of data compliance on large language model (LLM) performance. However, little is known about how these restrictions (and the resultant filtering of pretraining datasets) affect the capabilities of models trained using these corpora. In this work, we conceptualize this effect as the $\textit{data compliance gap}$ (DCG), which quantifies the performance difference between models trained on datasets that comply with web crawling opt-outs, and those that do not. We measure the data compliance gap in two settings: pretraining models from scratch and continual pretraining from existing compliant models (simulating a setting where copyrighted data could be integrated later in pretraining). Our experiments with 1.5B models show that, as of January 2025, compliance with web data opt-outs does not degrade general knowledge acquisition (close to 0\% DCG). However, in specialized domains such as biomedical research, excluding major publishers leads to performance declines. These findings suggest that while general-purpose LLMs can be trained to perform equally well using fully open data, performance in specialized domains may benefit from access to high-quality copyrighted sources later in training. Our study provides empirical insights into the long-debated trade-off between data compliance and downstream model performance, informing future discussions on AI training practices and policy decisions.
Abstract:As large language models increasingly rely on external data sources, fairly compensating data contributors has become a central concern. In this paper, we revisit the design of data markets through a game-theoretic lens, where data owners face private, heterogeneous costs for data sharing. We show that commonly used valuation methods--such as Leave-One-Out and Data Shapley--fail to ensure truthful reporting of these costs, leading to inefficient market outcomes. To address this, we adapt well-established payment rules from mechanism design, namely Myerson and Vickrey-Clarke-Groves (VCG), to the data market setting. We demonstrate that the Myerson payment is the minimal truthful payment mechanism, optimal from the buyer's perspective, and that VCG and Myerson payments coincide in unconstrained allocation settings. Our findings highlight the importance of incorporating incentive compatibility into data valuation, paving the way for more robust and efficient data markets.
Abstract:We target on-device collaborative fine-tuning of Large Language Models (LLMs) by adapting a Mixture of Experts (MoE) architecture, where experts are Low-Rank Adaptation (LoRA) modules. In conventional MoE approaches, experts develop into specialists throughout training. In contrast, we propose a novel $\textbf{Co}$llaborative learning approach via a $\textbf{Mi}$xture of $\textbf{G}$eneralists and $\textbf{S}$pecialists (CoMiGS). Diversifying into the two roles is achieved by aggregating certain experts globally while keeping others localized to specialize in user-specific datasets. Central to our work is a learnable routing network that routes at a token level, balancing collaboration and personalization at the finest granularity. Our method consistently demonstrates superior performance in scenarios with high data heterogeneity across various datasets. By design, our approach accommodates varying computational resource constraints among users as shown in different numbers of LoRA experts. We further showcase that low-resourced users can benefit from high-resourced users with high data quantity.
Abstract:We explore on-device self-supervised collaborative fine-tuning of large language models with limited local data availability. Taking inspiration from the collaborative learning community, we introduce three distinct trust-weighted gradient aggregation schemes: weight similarity-based, prediction similarity-based and validation performance-based. To minimize communication overhead, we integrate Low-Rank Adaptation (LoRA) and only exchange LoRA weight updates. Our protocols, driven by prediction and performance metrics, surpass both FedAvg and local fine-tuning methods, which is particularly evident in realistic scenarios with more diverse local data distributions. The results underscore the effectiveness of our approach in addressing heterogeneity and scarcity within local datasets.
Abstract:In this study, we systematically evaluate the impact of common design choices in Mixture of Experts (MoEs) on validation performance, uncovering distinct influences at token and sequence levels. We also present empirical evidence showing comparable performance between a learned router and a frozen, randomly initialized router, suggesting that learned routing may not be essential. Our study further reveals that Sequence-level routing can result in topic-specific weak expert specialization, in contrast to syntax specialization observed with Token-level routing.
Abstract:We consider a collaborative learning setting where each agent's goal is to improve their own model by leveraging the expertise of collaborators, in addition to their own training data. To facilitate the exchange of expertise among agents, we propose a distillation-based method leveraging unlabeled auxiliary data, which is pseudo-labeled by the collective. Central to our method is a trust weighting scheme which serves to adaptively weigh the influence of each collaborator on the pseudo-labels until a consensus on how to label the auxiliary data is reached. We demonstrate that our collaboration scheme is able to significantly boost individual model's performance with respect to the global distribution, compared to local training. At the same time, the adaptive trust weights can effectively identify and mitigate the negative impact of bad models on the collective. We find that our method is particularly effective in the presence of heterogeneity among individual agents, both in terms of training data as well as model architectures.
Abstract:Batch Normalization (BN) is widely used to stabilize the optimization process and improve the test performance of deep neural networks. The regularization effect of BN depends on the batch size and explicitly using smaller batch sizes with Batch Normalization, a method known as Ghost Batch Normalization (GBN), has been found to improve generalization in many settings. We investigate the effectiveness of GBN by disentangling the induced "Ghost Noise" from normalization and quantitatively analyzing the distribution of noise as well as its impact on model performance. Inspired by our analysis, we propose a new regularization technique called Ghost Noise Injection (GNI) that imitates the noise in GBN without incurring the detrimental train-test discrepancy effects of small batch training. We experimentally show that GNI can provide a greater generalization benefit than GBN. Ghost Noise Injection can also be beneficial in otherwise non-noisy settings such as layer-normalized networks, providing additional evidence of the usefulness of Ghost Noise in Batch Normalization as a regularizer.