作者位次 #2/7ACL 2026Poster· Natural Language Processing
D ec IF : Improving Instruction-Following through Decomposition
Tingfeng Hui · Pengyu Zhu · Bowen Ping · Ling Tang · Guanting Dong · Yaqi Zhang · Sen Su
摘要
We propose a novel data synthesis framework, DecIF, which automatically generates accurate and diverse instruction-following data from scratch for supervised fine-tuning (SFT) and reinforcement learning (RL), leveraging large language models (LLMs) and minimal external resources. By decomposing t...