Training Transformer language models is expensive, as performance typically improves with increasing dataset size and computational budget. Although scaling laws describe this trend at large scale, their implications in controlled, smaller-scale settings remain less explored. In this work, we isolate dataset-size effects using a strongly reduced attention-only decoder architecture. By training on progressively larger power-of-two subsets, we observe smooth performance improvements accompanied by clear diminishing returns, consistent with scaling-law behavior. Using only about 30% of the training data is sufficient to reach approximately 90% of the full-data validation token-level accuracy. These results provide actionable insights into dataset scaling in a controlled, component-isolated setting and offer practical guidance for balancing dataset size and computational cost in compute- and data-restricted environments, such as small research labs and exploratory model development.
@article{arxiv.2604.09389,
title = {Is More Data Worth the Cost? Dataset Scaling Laws in a Tiny Attention-Only Decoder},
author = {Götz-Henrik Wiegand and Lorena Raichle and Rico Städeli and Tomas Hrycej and Bernhard Bermeitinger and Siegfried Handschuh},
journal= {arXiv preprint arXiv:2604.09389},
year = {2026}
}
Comments
Presented as a paper at 3rd DATA-FM workshop @ ICLR 2026, Brazil. Published at 13th IEEE Swiss Conference on Data Science and AI (SDS 2026)