When AI Models Become Dependencies: Studying the Evolution of Pre-Trained Model Reuse in Downstream Software Systems

• URL: http://arxiv.org/abs/2604.17940v1
• Date: Mon, 20 Apr 2026 08:20:53 GMT
• Title: When AI Models Become Dependencies: Studying the Evolution of Pre-Trained Model Reuse in Downstream Software Systems
• Abstract summary: Modern software systems have transitioned from purely code-based architectures to AI-integrated systems where pre-trained models (PTMs) serve as permanent dependencies.<n>Unlike libraries, PTMs are characterized by opaque internals and less standardized, rapidly evolving release cycles.<n>This raises a critical question for software maintenance: do PTMs change like standard software libraries or do they follow a divergent pattern?<n>We present the first empirical study of downstream PTM changes, analyzing a comprehensive dataset of 4,988 releases across 323 GitHub OSS repositories.
• Score: 11.477328199957356
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Modern software systems have transitioned from purely code-based architectures to AI-integrated systems where pre-trained models (PTMs) serve as permanent dependencies. However, while the evolution of traditional software libraries is well-documented, we lack a clear understanding of how these "PTM dependencies" change over time. Unlike libraries, PTMs are characterized by opaque internals and less standardized, rapidly evolving release cycles. Furthermore, their multi-role nature enables developers to treat individual instances of a single PTM as separate functional dependencies based on their specific downstream tasks. This raises a critical question for software maintenance: do PTMs change like standard software libraries or do they follow a divergent pattern? To answer this, we present the first empirical study of downstream PTM changes, analyzing a comprehensive dataset of 4,988 releases across 323 GitHub OSS repositories that reuse open-source PTMs. Using traditional software libraries as a baseline, we find that PTMs follow a qualitatively distinct pattern. PTMs are typically added late in the project life-cycle and tend to accumulate rather than be replaced as a project matures. Our findings show that PTM changes are three times less frequent (406 of 2,814 release transitions) than library changes. PTM changes are also less routinely documented, but more likely to carry explicit rationale. Unlike libraries, which evolve reactively, PTM evolution is proactively driven by capability expansion, with a unique documented rationale of PTM testing uncertainty. Our work calls for a rethinking of how PTMs are tracked and managed as dependencies in modern software engineering.

Related papers

• Software Dependencies 2.0: An Empirical Study of Reuse and Integration of Pre-Trained Models in Open-Source Projects [9.22889135297242]
  Pre-trained models (PTMs) are machine learning models that have been trained in advance, often on large-scale data, and can be reused for new tasks.<n>Software Dependencies 2.0 introduces a new class of software dependency, which we term Software Dependencies 2.0.
  arXiv  Detail & Related papers  (2025-09-07T15:00:22Z)
• How do Pre-Trained Models Support Software Engineering? An Empirical Study in Hugging Face [52.257764273141184]
  Open-Source Pre-Trained Models (PTMs) provide extensive resources for various Machine Learning (ML) tasks.<n>These resources lack a classification tailored to Software Engineering (SE) needs.<n>We derive a taxonomy encompassing 147 SE tasks and apply an SE-oriented classification to PTMs in a popular open-source ML repository, Hugging Face (HF)<n>We find that code generation is the most common SE task among PTMs, while requirements engineering and software design activities receive limited attention.
  arXiv  Detail & Related papers  (2025-06-03T15:51:17Z)
• Towards a Classification of Open-Source ML Models and Datasets for Software Engineering [52.257764273141184]
  Open-source Pre-Trained Models (PTMs) and datasets provide extensive resources for various Machine Learning (ML) tasks. These resources lack a classification tailored to Software Engineering (SE) needs. We apply an SE-oriented classification to PTMs and datasets on a popular open-source ML repository, Hugging Face (HF), and analyze the evolution of PTMs over time.
  arXiv  Detail & Related papers  (2024-11-14T18:52:05Z)
• UniTST: Effectively Modeling Inter-Series and Intra-Series Dependencies for Multivariate Time Series Forecasting [98.12558945781693]
  We propose a transformer-based model UniTST containing a unified attention mechanism on the flattened patch tokens. Although our proposed model employs a simple architecture, it offers compelling performance as shown in our experiments on several datasets for time series forecasting.
  arXiv  Detail & Related papers  (2024-06-07T14:39:28Z)
• PeaTMOSS: A Dataset and Initial Analysis of Pre-Trained Models in Open-Source Software [6.243303627949341]
  This paper presents the PeaTMOSS dataset, which comprises metadata for 281,638 PTMs and detailed snapshots for all PTMs. The dataset includes 44,337 mappings from 15,129 downstream GitHub repositories to the 2,530 PTMs they use. Our analysis provides the first summary statistics for the PTM supply chain, showing the trend of PTM development and common shortcomings of PTM package documentation.
  arXiv  Detail & Related papers  (2024-02-01T15:55:50Z)
• PeaTMOSS: Mining Pre-Trained Models in Open-Source Software [6.243303627949341]
  We present the PeaTMOSS dataset: Pre-Trained Models in Open-Source Software. PeaTMOSS has three parts: a snapshot of 281,638 PTMs, (2) 27,270 open-source software repositories that use PTMs, and (3) a mapping between PTMs and the projects that use them.
  arXiv  Detail & Related papers  (2023-10-05T15:58:45Z)
• Efficient Federated Prompt Tuning for Black-box Large Pre-trained Models [62.838689691468666]
  We propose Federated Black-Box Prompt Tuning (Fed-BBPT) to optimally harness each local dataset. Fed-BBPT capitalizes on a central server that aids local users in collaboratively training a prompt generator through regular aggregation. Relative to extensive fine-tuning, Fed-BBPT proficiently sidesteps memory challenges tied to PTM storage and fine-tuning on local machines.
  arXiv  Detail & Related papers  (2023-10-04T19:30:49Z)
• Revisiting Class-Incremental Learning with Pre-Trained Models: Generalizability and Adaptivity are All You Need [84.3507610522086]
  Class-incremental learning (CIL) aims to adapt to emerging new classes without forgetting old ones. Recent pre-training has achieved substantial progress, making vast pre-trained models (PTMs) accessible for CIL. We argue that the core factors in CIL are adaptivity for model updating and generalizability for knowledge transferring.
  arXiv  Detail & Related papers  (2023-03-13T17:59:02Z)
• An Empirical Study of Pre-Trained Model Reuse in the Hugging Face Deep Learning Model Registry [2.1346819928536687]
  Machine learning engineers have begun to reuse large-scale pre-trained models (PTMs) We interviewed 12 practitioners from the most popular PTM ecosystem, Hugging Face, to learn the practices and challenges of PTM reuse. Three challenges for PTM reuse are missing attributes, discrepancies between claimed and actual performance, and model risks.
  arXiv  Detail & Related papers  (2023-03-05T02:28:15Z)
• Ranking and Tuning Pre-trained Models: A New Paradigm of Exploiting Model Hubs [136.4492678691406]
  We propose a new paradigm of exploiting model hubs by ranking and tuning pre-trained models. The best ranked PTM can be fine-tuned and deployed if we have no preference for the model's architecture. The tuning part introduces a novel method for multiple PTMs tuning, which surpasses dedicated methods.
  arXiv  Detail & Related papers  (2021-10-20T12:59:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.