Paper Archive

输入关键词搜索 102 篇论文

tracked author

Gennady Pekhimenko

LoRAFusion senior coauthor. University of Toronto homepage presents Gennady Pekhimenko as professor and EcoSystem group lead; public profiles connect him to CentML. X handle Dr_GGP matches name, role, and CentML/UofT systems identity.

1 archived notes X: high HomepageX

Related Notes

按论文归档时间排序，展示该作者在本站已经出现的材料。

Jun 28, 2026

LoRAFusion: Efficient LoRA Fine Tuning for LLMs

LoRAFusion 把 LoRA 微调的效率问题拆成两层：kernel 层用 split graph fusion 减少 LoRA 分支对大激活张量的重复读写，调度层把共享 base model 的多个 LoRA adapter 合并训练并用分组、MILP / greedy packing 降低流水线气泡和变长样本负载不均；在 H100 / L40S、LLaMA 3.1 8B、Qwen 2.5 32B、LLaMA 3.1 70B 上...

2510.00206-lorafusion-efficient-lora-fine-tuning Parameter-Efficient FinetuningDistributed TrainingTraining Memory