作者位次 #4/7ACL 2026Findings· Findings of ACL
D ynamix SFT : Dynamic Mixture Optimization of Instruction Tuning Collections
Haebin Shin · Lei Ji · Xiao Liu · Zhiwei Yu · Hyunwoo Yoo · Qi Chen · Yeyun Gong
摘要
As numerous instruction-tuning datasets continue to emerge, dynamically balancing and optimizing their mixtures has become a criticalchallenge. To address this, we propose DynamixSFT, a dynamic and automated method for instruction-tuning dataset mixture optimization. We formulate the problem as a...