用进化算法优化大模型，不用梯度也能训练0.5B到32B参数模型
EA4LLM: A Gradient-Free Approach to Large Language Model Optimization via Evolutionary Algorithms
- 采用无梯度的进化算法替代传统优化器，突破对可导性的依赖
- 首次在0.5B至32B参数规模上实现从预训练开始的全参数优化
- 适合算力有限的研究者，降低大模型训练的硬件门槛
近年来，大语言模型（LLMs）取得显著进展，但其优化主要依赖于梯度方法（如Adam），对硬件要求高，需高并发、高内存的GPU。同时，这些方法要求所有神经网络操作可微，导致许多非可导架构无法实际应用。为此，我们提出EA4LLM，一种用于优化大语言模型的进化算法，并首次在0.5B至32B参数规模范围内，实证验证了从预训练阶段开始的全参数优化。通过大量实验，揭示了进化算法有效优化神经网络的关键机制。本工作挑战了梯度优化是训练神经网络唯一可行方法的普遍认知，具有显著降低大模型训练计算成本的潜力，使资源有限的研究团队也能参与深度学习研究。
原文摘要 · Abstract (English)
In recent years, large language models (LLMs) have made remarkable progress, with model optimization primarily relying on gradient-based optimizers such as Adam. However, these gradient-based methods impose stringent hardware requirements, demanding high-concurrency, high-memory GPUs. Moreover, they require all neural network operations to be differentiable, thereby excluding many promising non-differentiable architectures from practical use. To address these limitations, we propose EA4LLM, an evolutionary algorithm for optimizing LLMs, and, for the first time, empirically verify full-parameter optimization from the pretraining stage across model sizes ranging from 0.5B to 32B. We conduct extensive experiments and provide key insights into how evolutionary algorithms can effectively optimize neural networks. Our work challenges the prevailing assumption that gradient-based optimization is the only viable approach for training neural networks. It also holds significant potential to reduce the computational cost of training large language models, thereby enabling groups with limited computational resources to participate in deep learning research.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。