Mini-Sequence Transformers: Optimizing Intermediate Memory for Long Sequences Training

We introduce Mini-Sequence Transformer (MsT), a simple and effective methodology for highly efficient and accurate LLM training with extremely long sequences. MsT partitions input sequences and iteratively processes mini-sequences to reduce intermediate memory usage. Integrated with activation recomputation, it enables significant memory savings in both forward and backward passes. In experiments with the Llama3-8B model, with MsT, we measure no degradation in throughput or convergence even with 12x longer sequences than standard implementations. MsT is fully general, implementation-agnostic, and requires minimal code changes to integrate with existing LLM training frameworks. Integrated with the huggingface library, MsT successfully extends the maximum context length of Qwen, Mistral, and Gemma-2 by 12-24x.

Authors

• Cheng Luo
• Jiawei Zhao
• Zhuoming Chen
• Beidi Chen
• Anima Anandkumar

Keywords

No keywords are indexed for this paper.