Is More Data Worth the Cost? Dataset Scaling Laws in a Tiny Attention-Only Decoder

Götz-Henrik Wiegand, Lorena Raichle, Rico Staedeli, Tomas Hrycej, Bernhard Bermeitinger, Siegfried Handschuh

Keywords: dataset scaling, attention-only Transformers, data efficiency, scaling laws, compute-efficient pretraining, large language models

TL;DR: Controlled experiments with tiny attention-only decoders demonstrate that dataset scaling laws persist at small scales, where training on just 30% of the data achieves 90% of the full performance.

Abstract:

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.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 61