Predicting Task Performance with Context-aware Scaling Laws

Kyle Montgomery, David Park, Jianhong Tu, Michael Bendersky, Beliz Gunel, Dawn Song, Chenguang Wang

Keywords: scaling, generalization, long-context

TL;DR: This paper introduces a context-aware scaling framework that jointly models downstream performance from training compute and context length, and it demonstrates its accuracy and generalizability across several tasks.

Abstract: Scaling laws have transformed our understanding of large language models by linking upstream metrics like cross-entropy loss to design factors such as model size, training data, and compute. However, these conventional laws fail to capture downstream task performance, where context plays a critical role. In this work, we propose a straightforward, interpretable framework that jointly models downstream performance as a function of the training compute and the provided context. We empirically validate our framework by fitting it on the observed downstream performance of extended-context variants of Llama-2-7B and Llama-2-13B across 65,500 unique instances spanning three tasks: arithmetic reasoning, common sense reasoning, and machine translation. Our results demonstrate that our framework accurately models in-distribution downstream performance, generalizes across three orders of magnitude in training compute, and reliably extrapolates performance as the amount of context increases. These findings offer valuable insights into the interplay between training compute and context utilization, providing guidance for designing more efficient long-context LLMs for diverse downstream tasks. Our code is available at https://github.com/wang-research-lab/context-scaling.

Archival Status: Non-archival (not included in proceedings)

Submission Number: 36