Soft Stopping Conditions for Long Training Runs
Aravind Periyasamy, Liora Sabatini, Marek Holub, Karima Belkadi
@inproceedings{periyasamy2025softstop,
title = {Soft Stopping Conditions for Long Training Runs},
author = {Periyasamy, Aravind and Sabatini, Liora and Holub, Marek and Belkadi, Karima},
year = {2025},
booktitle = {alphabell index 25/20 · ML Safety Workshop, NeurIPS 2025},
month = {dec},
doi = {10.48550/arXiv.2512.07221},
url = {https://dev.alphabell.com/publications/soft-stopping-conditions-long-runs}
}
Abstract
Hard stopping conditions — those that halt a run the moment a tripwire is crossed — are appropriate for short or moderate-length runs but produce expensive false-positive halts on training runs that span multiple weeks. We introduce soft stopping conditions: a calibrated, signed-by-paired-cell escalation protocol in which an intermediate trip raises a flag rather than halts the run, while preserving the option of subsequent escalation. We discuss when soft stops are appropriate, when they are not, and how the MUR protocol is amended to support them.
Index metadata
- Cell
- godel-02
- Compute
- redacted
- Status
- Open release with companion redacted appendix
- Companion
- MUR protocol 25/05 v2 amendment
- DOI
- 10.48550/arXiv.2512.07221
- arXiv
- arXiv:2512.07221
What this paper is part of
This index entry is part of the Recursive self-improvement research axis. The producing cell — godel-02 — collaborates with adjacent cells listed in the cell directory. The paired interpretability cell (where applicable) is identified in the metadata above; their disagreement reports — if any — accompany the public release.
How to read this
If you want to use the result: the code (where available) is at https://github.com/alphabell-labs/ab-soft-stop; the dataset is at TBD when one is released. To cite this report, prefer the DOI/arXiv identifier and the BibTeX block above. To discuss this with the producing cell, contact the lab with the index entry slug soft-stopping-conditions-long-runs.
Limitations
Each cell-published report carries an explicit limitations section in the internal index. We do not paraphrase it here. Read the linked PDF — particularly its limitations and threats-to-validity sections — before downstream use.
Aravind Periyasamy, Liora Sabatini, Marek Holub, Karima Belkadi. Soft Stopping Conditions for Long Training Runs. alphabell index 25/20 · ML Safety Workshop, NeurIPS 2025, Dec 2025. arXiv:2512.07221. doi:10.48550/arXiv.2512.07221.