Nada Zine, Matías Martínez, Romain Rouvoy, Clément Quinton · 2026 · arXiv (Cornell University)
DOI: 10.48550/arxiv.2610.27945 · cited by 0 works (OpenAlex) · open-access PDF
En este artículo, presentamos un estudio empírico a gran escala de las prácticas de configuración de la inferencia de LLM en software de código abierto utilizando Hugging Face Transformers, vLLM y la API OpenAI. Analizamos 5,398 repositorios optimizados de GitHub utilizando al menos uno de estos backends e identificamos 345,117 ocurrencias de opciones explicitamente configuradas. Nuestros resultados muestran que los desarrolladores utilizan solo una pequeña parte del espacio de configuración disponible regularmente. Aunque el 98,8% de las opciones de configuración de fondo documentadas aparecen al menos una vez, el 78,9% de los repositorios están configurados en la mayoría de los 10% de los repositorios para los estudios de reimportación de sustancias correspondientes. Las observaciones de las opciones de reimportación a través de las opciones de reimportación
Machine translation (Meta M2M100); the English original follows.
LLMs are increasingly integrated into software systems, making deployment and inference configuration a recurring software engineering task. Inference backends expose many options that affect model loading, generation, serving, and request processing. Yet little is known about which of these options developers explicitly configure in practice. In this paper, we present a large-scale empirical study of LLM inference configuration practices in open-source software using Hugging Face Transformers, vLLM, and the OpenAI API. We analyze 5,398 GitHub repositories using at least one of these backends and identify 345,117 explicitly configured option occurrences. Our results show that developers use only a small part of the available configuration space regularly. Although 98.8% of the documented backend-specific options appear at least once, 78.9% are configured in at most 10% of repositories using the corresponding backend. Configurations are also small, with a median of 2 options for Transformers and 4 for vLLM and OpenAI, but the combinations of options vary widely: 56.8-61.6% of multi-option signatures appear in only one repository. We also observe recurring option values as well as substantial variation in how values are chosen. Finally, configuration practices differ across application contexts, with option-adoption differences reaching 30.7 percentage points. These findings characterize the configuration surface developers make explicit rather than the importance or optimality of individual options. They provide empirical evidence for prioritizing future studies of inference configuration and for improving backend documentation and configuration support.
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
Topics: Software deployment, Documentation, Inference, Computer science, Software engineering, Software
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.27945, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange