{"ID":2857454,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.09180","arxiv_id":"2510.09180","title":"RepDL: Bit-level Reproducible Deep Learning Training and Inference","abstract":"Non-determinism and non-reproducibility present significant challenges in deep learning, leading to inconsistent results across runs and platforms. These issues stem from two origins: random number generation and floating-point computation. While randomness can be controlled through deterministic configurations, floating-point inconsistencies remain largely unresolved. To address this, we introduce RepDL, an open-source library that ensures deterministic and bitwise-reproducible deep learning training and inference across diverse computing environments. RepDL achieves this by enforcing correct rounding and order invariance in floating-point computation. The source code is available at https://github.com/microsoft/RepDL .","short_abstract":"Non-determinism and non-reproducibility present significant challenges in deep learning, leading to inconsistent results across runs and platforms. These issues stem from two origins: random number generation and floating-point computation. While randomness can be controlled through deterministic configurations, floati...","url_abs":"https://arxiv.org/abs/2510.09180","url_pdf":"https://arxiv.org/pdf/2510.09180v1","authors":"[\"Peichen Xie\",\"Xian Zhang\",\"Shuo Chen\"]","published":"2025-10-10T09:24:07Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.SE\"]","methods":"[]","has_code":false,"code_links":[{"ID":608453,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2857454,"paper_url":"https://arxiv.org/abs/2510.09180","paper_title":"RepDL: Bit-level Reproducible Deep Learning Training and Inference","repo_url":"https://github.com/microsoft/RepDL","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}