arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2412.06540cs.LGcs.AIstat.ML
收藏

Sloth: LLM技能的缩放定律用于跨家族预测多基准性能

Sloth: scaling laws for LLM skills to predict multi-benchmark performance across families

Department of Statistics, University of Michigan(密歇根大学统计学系);MIT-IBM Watson AI Lab, IBM Research(MIT-IBM Watson AI实验室,IBM研究);Computer Science and Artificial Intelligence Laboratory, MIT(MIT计算机科学与人工智能实验室);Institute of Foundation Models, MBZUAI(基础模型研究所,MBZUAI)

Felipe Maia Polo, Seamus Somerstep, Leshem Choshen, Yuekai Sun, Mikhail Yurochkin

AI总结:

Sloth提出一种基于低维潜在技能的LLM缩放定律,通过跨基准相关性提升预测准确性,减少多家族训练需求。

摘要:

大型语言模型(LLM)的缩放定律根据参数如大小和训练数据预测模型性能。然而,不同模型家族在训练配置和数据处理上的差异导致基准性能显著变化,使得单一缩放定律难以跨所有LLM通用。另一方面,为每个家族训练特定的缩放定律需要为每个家族训练不同大小的模型。在本工作中,我们提出了Skills Scaling Laws(SSLaws,发音为Sloth),一种新的缩放定律,利用公开可用的基准数据,并假设LLM性能由低维潜在技能(如推理和指令遵循)驱动。这些潜在技能受计算资源(如模型大小和训练令牌)影响,但不同模型家族的效率各不相同。Sloth通过跨基准的相关性提供更准确和可解释的预测,同时减少为每个家族训练多个LLM的需要。我们展示了参数识别的理论结果和在12个 prominent 基准上的实证评估,从Open LLM Leaderboard v1/v2中得出,证明Sloth能准确预测LLM性能,并提供关于复杂下游任务、增加测试时间计算和计算最优技能缩放的见解。

英文摘要:

Scaling laws for large language models (LLMs) predict model performance based on parameters like size and training data. However, differences in training configurations and data processing across model families lead to significant variations in benchmark performance, making it difficult for a single scaling law to generalize across all LLMs. On the other hand, training family-specific scaling laws requires training models of varying sizes for every family. In this work, we propose Skills Scaling Laws (SSLaws, pronounced as Sloth), a novel scaling law that leverages publicly available benchmark data and assumes LLM performance is driven by low-dimensional latent skills, such as reasoning and instruction following. These latent skills are influenced by computational resources like model size and training tokens, but with varying efficiencies across model families. Sloth exploits correlations across benchmarks to provide more accurate and interpretable predictions while alleviating the need to train multiple LLMs per family. We present both theoretical results on parameter identification and empirical evaluations on 12 prominent benchmarks, from Open LLM Leaderboard v1/v2, demonstrating that Sloth predicts LLM performance accurately and offers insights into scaling behaviors for complex downstream tasks, increased test-time compute, and compute-optimal scaling of skills.

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。

↑