arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2401.00788cs.CLcs.AIcs.SE
收藏

Astraios:代码大型语言模型的参数高效指令微调

Astraios: Parameter-Efficient Instruction Tuning Code Large Language Models

Monash University(莫纳什大学);Hugging Face;ServiceNow Research(ServiceNow 研究院);Sea AI Lab(Sea AI 实验室);Contextual AI

Terry Yue Zhuo, Armel Zebaze, Nitchakarn Suppattarachai, Leandro von Werra, Harm de Vries, Qian Liu, Niklas Muennighoff

AI总结:

本文提出Astraios套件,包含28个指令微调OctoCoder模型,通过7种方法和4种规模评估发现FFT性能最佳,LoRA权衡最优,大模型鲁棒性与安全性下降,且小模型微调有效性可推广至大模型。

摘要:

大型语言模型(LLMs)全参数微调(FFT)的高成本催生了一系列参数高效微调(PEFT)方法。然而,目前尚不清楚在不同模型规模下,哪些方法能提供最佳的成本与性能权衡。我们引入了Astraios,这是一套包含28个指令微调OctoCoder模型的套件,使用了7种微调方法和4种高达160亿参数的模型规模。通过对涵盖代码理解和代码生成任务的5项任务和8个不同数据集进行调查,我们发现FFT通常在所有规模下都能带来最佳的下游性能,而PEFT方法的有效性因模型规模而异。LoRA通常提供最有利的成本与性能权衡。对这些方法在模型鲁棒性和代码安全性方面影响的进一步调查表明,较大的模型往往表现出鲁棒性下降和安全性降低。最后,我们探索了更新参数、交叉熵损失和任务性能之间的关系。我们发现,在小模型中观察到的微调有效性可以很好地推广到较大的模型中,并且指令微调中的验证损失可以作为整体下游性能的可靠指标。

英文摘要:

The high cost of full-parameter fine-tuning (FFT) of Large Language Models (LLMs) has led to a series of parameter-efficient fine-tuning (PEFT) methods. However, it remains unclear which methods provide the best cost-performance trade-off at different model scales. We introduce Astraios, a suite of 28 instruction-tuned OctoCoder models using 7 tuning methods and 4 model sizes up to 16 billion parameters. Through investigations across 5 tasks and 8 different datasets encompassing both code comprehension and code generation tasks, we find that FFT generally leads to the best downstream performance across all scales, and PEFT methods differ significantly in their efficacy based on the model scale. LoRA usually offers the most favorable trade-off between cost and performance. Further investigation into the effects of these methods on both model robustness and code security reveals that larger models tend to demonstrate reduced robustness and less security. At last, we explore the relationships among updated parameters, cross-entropy loss, and task performance. We find that the tuning effectiveness observed in small models generalizes well to larger models, and the validation loss in instruction tuning can be a reliable indicator of overall downstream performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑