Metraipolis › archive › dc_7afddb5ecf9e
Predicting Deep Neural Network Training Outcomes from Early Training Telemetry — arXiv:2608.03709
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.03709
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T12:31:10.276Z
- SHA-256 of the archived text
bb4897e508c044711ed7b9974beb038caa0b79c3e8c95aa0005bf26e74980249- Stable URL of this page
https://metraipolis.com/documento/dc_7afddb5ecf9e
How to cite
Predicting Deep Neural Network Training Outcomes from Early Training Telemetry — arXiv:2608.03709 — arXiv (en), https://arxiv.org/abs/2608.03709 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7afddb5ecf9e (Metraipolis archive dc_7afddb5ecf9e, sha256 bb4897e508c044711ed7b9974beb038caa0b79c3e8c95aa0005bf26e74980249)
BibTeX
@misc{dc7afddb5ecf9e,
title = {Predicting Deep Neural Network Training Outcomes from Early Training Telemetry — arXiv:2608.03709},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.03709},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7afddb5ecf9e; sha256 bb4897e508c044711ed7b9974beb038caa0b79c3e8c95aa0005bf26e74980249},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.03709. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large hyperparameter sweeps for deep neural networks spend substantial compute on configurations that are effectively doomed from the first few epochs. We study whether a single training run's own early telemetry - per-epoch loss, training accuracy, gradient signal-to-noise ratio, weight-norm growth, and an activation-saturation snapshot - together with its sampled hyperparameters, can predict that run's eventual outcome without reference to other runs. We evaluate three prediction tasks: final test accuracy, relative performance within a domain, and training-dynamics failure, including numerical divergence. Across 23,788 training runs spanning six architecture/dataset combinations, gradient-boosted trees using only the first five epochs of telemetry achieve R^2 = 0.92-0.99 for final-accuracy regression and ROC-AUC = 0.983-0.998 for relative classification on a permanently held-out set of hyperparameter configurations. Useful prediction is already available after a single epoch. A paired ablation shows that gradient- and weight-level telemetry provides a statistically consistent improvement over loss and accuracy curves alone, although the practical gain varies by domain. Transfer is strong between similar architectures, while cross-dataset transfer is limited mainly by differences in accuracy scale rather than loss of the underlying relationship. These results suggest that early-training telemetry can provide a practical decision-support signal for compute allocation while motivating human oversight for any automated intervention. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2608.03709 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-31T12:17:37.777Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).