SkyLadder: Better and Faster Pretraining via Context Window Scheduling

Tongyao Zhu, Qian Liu, Haonan Wang, Shiqi Chen, Xiangming Gu, Tianyu Pang, Min-Yen Kan

Keywords: pretraining, language models, context window

TL;DR: We investigate an efficient and effective context-window scheduling method for language model pretraining.

Abstract: Recent advancements in LLM pretraining have featured ever-expanding context windows to process longer sequences. However, our controlled study reveals that models pretrained with shorter context windows consistently outperform their long-context counterparts under a fixed token budget. This finding motivates us to explore an optimal context window scheduling strategy to better balance long-context capability with pretraining efficiency. To this end, we propose SkyLadder, a simple yet effective approach that implements a short-to-long context window transition. SkyLadder preserves strong standard benchmark performance, while matching or exceeding baseline results on long-context tasks. Through extensive experiments, we pretrain 1B-parameter models (up to 32K context) and 3B-parameter models (8K context) on 100B tokens, demonstrating that SkyLadder yields consistent gains of up to 3.7% on common benchmarks, while achieving up to 22% faster training speeds compared to baselines.

Primary Area: Deep learning (e.g., architectures, generative models, optimization for deep networks, foundation models, LLMs)

Submission Number: 6948