想省训练算力又不想掉性能？这个嵌套训练法把三个模型塞一个架构里，直接省36%算力，还顺带加速推理，值得一看。
Matryoshka训练框架将500M、1.5B和3B三个子模型堆叠进单一嵌套架构，端到端训练。相比独立训练基线，该套件在基准性能和困惑度上持平，但训练计算量减少36%。嵌套结构天然支持投机解码，吞吐量提升14-26%。论文还消融了关键架构选择，为构建强Matryoshka套件提供指导。
Matryoshka Language Model Suites
Training a language model suite classically requires training each model separately and serving them independently. We improve both training and inference efficiency by stacking sub-models of increasing size into a single nested architecture trained end-to-end. This Matryoshka training framework reduces the total parameter count of the suite, enables low-cost distillation from the largest to all smaller sub-models at every training step, and is well-suited for speculative decoding as the draft model is contained within the verifier. We validate our approach by training a Matryoshka suite comprising 500M, 1.5B, and 3B sub-models. Our suite is on par with independently trained baselines on benchmark performance and validation and out-of-domain perplexities, while using 36% less training compute and improving the throughput of speculative decoding by 14-26%. We also ablate key architectural choices, offering guidance for building strong Matryoshka LM suites.