5 papers
CODEBLOCK: Learning to Supervise Code at the Right Granularity
Zhijie Deng, Ling Li, Jinlong Pang +4
Supervised fine-tuning of code LLMs typically applies uniform cross-entropy loss to all response tokens, implicitly assuming that every token provides equally useful learning signa…
Token Cleaning: Fine-Grained Data Selection for LLM Supervised Fine-Tuning
Jinlong Pang, Na Di, Zhaowei Zhu +4
Recent studies show that in supervised fine-tuning (SFT) of large language models (LLMs), data quality matters more than quantity. While most data cleaning methods concentrate on f…
Improving Data Efficiency via Curating LLM-Driven Rating Systems
Jinlong Pang, Jiaheng Wei, Ankit Parag Shah +6
Instruction tuning is critical for adapting large language models (LLMs) to downstream tasks, and recent studies have demonstrated that small amounts of human-curated data can outp…
Reassessing Layer Pruning in LLMs: New Insights and Methods
Yao Lu, Hao Cheng, Yujie Fang +6
Although large language models (LLMs) have achieved remarkable success across various domains, their considerable scale necessitates substantial computational resources, posing sig…
LLM Unlearning via Loss Adjustment with Only Forget Data
Yaxuan Wang, Jiaheng Wei, Chris Yuhao Liu +6
Unlearning in Large Language Models (LLMs) is essential for ensuring ethical and responsible AI use, especially in addressing privacy leak, bias, safety, and evolving regulations.…