An Empirical Study of Artifacts and Security Risks in the Pre-trained Model Supply Chain

Jiang, Wenxin; Synovic, Nicholas M.; Läufer, Konstantin; Indarapu, Aryan; Hyatt, Matt; Schorlemmer, Taylor R.; Thiruvathukal, George K.; Davis, James N.

doi:10.1145/3560835.3564547

Cited by 17 publications

(15 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similarly, Vu et al highlighted existing discrepancies at different levels of granularity in PyPi [30]. Following the machine learning scientific research community [31], the software engineering community has just begun to study concerns in DL model registries [32]. We offer an early software engineering view on this topic.…”

Section: Background and Related Workmentioning

confidence: 97%

“…We studied the reusability of PTM packages in DL model registries, examining qualitative and quantitative aspects. We focused on one DL model registry, Hugging Face, as it is by far the largest registry at present [19]. For PTM reuse in the Hugging Face ecosystem, we ask: RQ1 How do engineers select PTMs?…”

Section: Research Questionsmentioning

confidence: 99%

“…Our interview study follows a four-step process modeled on the framework analysis methodology [64,65] (1) Data Familiarization and Framework Identification Our initial thematic framework is based on three themes from our literature review ( §II): model selection, PTM attributes, and PTM trustworthiness. For model selection, the identified considerations were the PTM reuse issues and factors affecting the decision-making process [19,66]. For attributes, we saw both traditional attributes (i.e., popularity, quality, maintenance) [67,68], and DL-specific attributes, viz.…”

Section: A Qualitative Study: Interviews With Ptm Reusersmentioning

confidence: 99%

“…provenance, reproducibility, and portability [52][53][54], shown in the first three columns in Table I. For trustworthiness, we considered the aspects assumed trustworthy plus possible discrepancies [19].…”

Section: A Qualitative Study: Interviews With Ptm Reusersmentioning

confidence: 99%

“…We took a mixed-methods approach to identify diverse phenomena for future investigation [18]. We focused our study on the Hugging Face DL model registry, which is the largest PTM registry at present [19]. First, we interviewed 12 Hugging Face practitioners to understand the practices and challenges of PTM reuse.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registry

Jiang¹,

Synovic²,

Hyatt³

et al. 2023

Preprint

View full text Add to dashboard Cite

Deep Neural Networks (DNNs) are being adopted as components in software systems. Creating and specializing DNNs from scratch has grown increasingly difficult as stateof-the-art architectures grow more complex. Following the path of traditional software engineering, machine learning engineers have begun to reuse large-scale pre-trained models (PTMs) and fine-tune these models for downstream tasks. Prior works have studied reuse practices for traditional software packages to guide software engineers towards better package maintenance and dependency management. We lack a similar foundation of knowledge to guide behaviors in pre-trained model ecosystems.In this work, we present the first empirical investigation of PTM reuse. We interviewed 12 practitioners from the most popular PTM ecosystem, Hugging Face, to learn the practices and challenges of PTM reuse. From this data, we model the decision-making process for PTM reuse. Based on the identified practices, we describe useful attributes for model reuse, including provenance, reproducibility, and portability. Three challenges for PTM reuse are missing attributes, discrepancies between claimed and actual performance, and model risks. We substantiate these identified challenges with systematic measurements in the Hugging Face ecosystem. Our work informs future directions on optimizing deep learning ecosystems by automated measuring useful attributes and potential attacks, and envision future research on infrastructure and standardization for model registries.

show abstract

Section: Background and Related Workmentioning

confidence: 97%

Section: Research Questionsmentioning

confidence: 99%

Section: A Qualitative Study: Interviews With Ptm Reusersmentioning

confidence: 99%

Section: A Qualitative Study: Interviews With Ptm Reusersmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registry

Jiang¹,

Synovic²,

Hyatt³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

Contrastive Knowledge Amalgamation for Unsupervised Image Classification

Gao,

Fu,

Liu

et al. 2023

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Challenges and practices of deep learning model reengineering: A case study on computer vision

Jiang,

Banna,

Vivek

et al. 2024

Empir Software Eng

View full text Add to dashboard Cite

Context Many engineering organizations are reimplementing and extending deep neural networks from the research community. We describe this process as deep learning model reengineering. Deep learning model reengineering — reusing, replicating, adapting, and enhancing state-of-the-art deep learning approaches — is challenging for reasons including under-documented reference models, changing requirements, and the cost of implementation and testing. Objective Prior work has characterized the challenges of deep learning model development, but as yet we know little about the deep learning model reengineering process and its common challenges. Prior work has examined DL systems from a “product” view, examining defects from projects regardless of the engineers’ purpose. Our study is focused on reengineering activities from a “process” view, and focuses on engineers specifically engaged in the reengineering process. Method Our goal is to understand the characteristics and challenges of deep learning model reengineering. We conducted a mixed-methods case study of this phenomenon, focusing on the context of computer vision. Our results draw from two data sources: defects reported in open-source reeengineering projects, and interviews conducted with practitioners and the leaders of a reengineering team. From the defect data source, we analyzed 348 defects from 27 open-source deep learning projects. Meanwhile, our reengineering team replicated 7 deep learning models over two years; we interviewed 2 open-source contributors, 4 practitioners, and 6 reengineering team leaders to understand their experiences. Results Our results describe how deep learning-based computer vision techniques are reengineered, quantitatively analyze the distribution of defects in this process, and qualitatively discuss challenges and practices. We found that most defects (58%) are reported by re-users, and that reproducibility-related defects tend to be discovered during training (68% of them are). Our analysis shows that most environment defects (88%) are interface defects, and most environment defects (46%) are caused by API defects. We found that training defects have diverse symptoms and root causes. We identified four main challenges in the DL reengineering process: model operationalization, performance debugging, portability of DL operations, and customized data pipeline. Integrating our quantitative and qualitative data, we propose a novel reengineering workflow. Conclusions Our findings inform several conclusion, including: standardizing model reengineering practices, developing validation tools to support model reengineering, automated support beyond manual model reengineering, and measuring additional unknown aspects of model reengineering.

show abstract

An Empirical Study of Artifacts and Security Risks in the Pre-trained Model Supply Chain

Cited by 17 publications

References 33 publications

An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registry

An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registry

Contrastive Knowledge Amalgamation for Unsupervised Image Classification

Challenges and practices of deep learning model reengineering: A case study on computer vision

Contact Info

Product

Resources

About