Observational Scaling Laws and the Predictability of Language Model Performance

arXiv:2405.10938v1 Announce Type: new
Abstract: Understanding how language model performance varies with scale is critical to benchmark and algorithm development. Scaling laws are one approach to building this understanding, but the requirement of training models across many different scales has limited their use. We propose an alternative, observational approach that bypasses model training and instead builds scaling laws from ~80 publically available models. Building a single scaling law from multiple model families is challenging due to large variations in their …

abstract algorithm alternative arxiv benchmark building cs.ai cs.cl cs.lg development language language model laws performance scale scaling stat.ml training training models type understanding