Research map / Architectures and efficiency / RNNs and normalization
Fast Catch-Up, Late Switching: Optimal Batch Size Scheduling via Functional Scaling Laws
Jinbo Wang, Binghui Li, Zhanpeng Zhou, Mingze Wang, Yuxuan Sun, Jiaqi Zhang, Xunliang Cai, Lei Wu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| defer the switch from small to large batch to a late stage (about 70% of tokens) instead of early (ablation) | two-stage batch size schedule · 1B MoE, 0.4T tokens (C4/LLaMA small scale also) | validation loss | Improved |
| late-switch schedule vs constant large batch and early-switch baselines (ablation) | Dense and MoE pretraining up to 1.1B params, 1T tokens · LLM pretraining | validation loss | Improved |
| switch batch size from small to large mid-training | constant small-batch training · LLM pretraining (fast catch-up) | validation loss vs large-batch trajectory | Improved |
Closest papers by meaning
- How Does Critical Batch Size Scale in Pre-training? (ICLR 2025 · similarity 0.89)
- Power Lines: Scaling laws for weight decay and batch size in LLM pre-training (NeurIPS 2025 · similarity 0.89)
- Functional Scaling Laws in Kernel Regression: Loss Dynamics and Learning Rate Schedules (NeurIPS 2025 · similarity 0.88)
- Critical Batch Size Revisited: A Simple Empirical Approach to Large-Batch Language Model Training (NeurIPS 2025 · similarity 0.88)
- Resolving Discrepancies in Compute-Optimal Scaling of Language Models (NeurIPS 2024 · similarity 0.88)
- Small Batch Size Training for Language Models: When Vanilla SGD Works, and Why Gradient Accumulation is Wasteful (NeurIPS 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
