用同家族模型指导LLM改进低性能模型，效果显著提升
Curating Same-Family Neural Networks for LLM-Guided Model Improvement: A Controlled Case Study
- 用已有模型作为提示，引导LLM优化目标模型
- CIFAR-10准确率从23.98%提升至50.49%，SVHN达78.80%
- 效果依赖源模型与目标的匹配度，非单纯看源模型精度
神经网络仓库包含可执行模型、训练配方、输入变换及测量精度。我们研究同一家族实验能否作为提示，指导基于大语言模型（LLM）的方法，在相同生成与评估预算下改进低性能目标模型。TuneNNGen在NNGPT基础上引入源模型引导路径，对比仅目标生成策略，在一个CIFAR-10目标、两个固定源选择规则、三种代码LLM及一个额外的SVHN目标上进行实验。在历史单轮搜索协议下，可用评估集上的最佳预算准确率在CIFAR-10上从23.98%提升至50.49%，在SVHN上从22.54%提升至78.80%。五轮、三种子的训练推导验证集平均表现仍保持提升：CIFAR-10提升40.94点，Imagenette提升18.83点，CIFAR-100提升7.27点。直接复制与负向控制分析表明，性能增益取决于源-目标兼容性与LLM适应能力，仅存储的源模型精度无法预测迁移效果。
原文摘要 · Abstract (English)
Neural-network repositories contain executable models, recipes, input transformations, and measured accuracies. We study whether one same-family experiment can be curated as prompt guidance for LLM-based improvement of a low-performing target under equal generation and evaluation budgets. TuneNNGen extends NNGPT with a source-guided route and compares it with target-only generation on one CIFAR-10 target, two fixed source-selection rules, three code LLMs, and an additional SVHN target. Under the historical one-epoch search protocol, best-of-budget accuracy on the available evaluation split rises from 23.98% to 50.49% on CIFAR-10 and from 22.54% to 78.80% on SVHN. Selected five-epoch, three-seed means on train-derived validation splits retain gains of 40.94 points on CIFAR-10, 18.83 on Imagenette, and 7.27 on CIFAR-100. Direct-copy and negative-control analyses show that gains depend on source-target compatibility and LLM adaptation; stored source accuracy alone does not predict transferability.
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。