# Exploring Learngene via Stage-wise Weight Sharing for Initializing Variable-sized Models

arXiv: https://arxiv.org/abs/2404.16897 · HTML: https://mlsanity.com/paper/2404.16897/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Builds on 31 of its 45 references that are in the map.
- First posted on arXiv 25 Apr 2024.

## Builds on (most cited first, 30 of 31)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [LLaMA: Open and Efficient Foundation Language Models](https://mlsanity.com/paper/2302.13971/) (2023)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification](https://mlsanity.com/paper/1502.01852/) (2015)
- [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://mlsanity.com/paper/1909.11942/) (2019)
- [BEiT: BERT Pre-Training of Image Transformers](https://mlsanity.com/paper/2106.08254/) (2021)
- [Deep Equilibrium Models](https://mlsanity.com/paper/1909.01377/) (2019)
- [Image as a Foreign Language: BEiT Pretraining for All Vision and Vision-Language Tasks](https://mlsanity.com/paper/2208.10442/) (2022)
- [LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inference](https://mlsanity.com/paper/2104.01136/) (2021)
- [Revealing the Dark Secrets of BERT](https://mlsanity.com/paper/1908.08593/) (2019)
- [All you need is a good init](https://mlsanity.com/paper/1511.06422/) (2015)
- [MiniViT: Compressing Vision Transformers with Weight Multiplexing](https://mlsanity.com/paper/2204.07154/) (2022)
- [Lessons on Parameter Sharing across Layers in Transformers](https://mlsanity.com/paper/2104.06022/) (2021)
- [Initializing Models with Larger Ones](https://mlsanity.com/paper/2311.18823/) (2023)
- [Recurrent Stacking of Layers for Compact Neural Machine Translation Models](https://mlsanity.com/paper/1807.05353/) (2018)
- [How to Initialize your Network? Robust Initialization for WeightNorm & ResNets](https://mlsanity.com/paper/1906.02341/) (2019)
- [Weight subcloning: direct initialization of transformers using larger pretrained ones](https://mlsanity.com/paper/2312.09299/) (2023)
- [Learngene: From Open-World to Your Learning Task](https://mlsanity.com/paper/2106.06788/) (2021)
- [TinyMIM: An Empirical Study of Distilling MIM Pre-trained Models](https://mlsanity.com/paper/2301.01296/) (2023)
- [Learngene: Inheriting Condensed Knowledge from the Ancestry Model to Descendant Models](https://mlsanity.com/paper/2305.02279/) (2023)
- [Transformer as Linear Expansion of Learngene](https://mlsanity.com/paper/2312.05614/) (2023)
- [Transferring Core Knowledge via Learngenes](https://mlsanity.com/paper/2401.08139/) (2024)

## Built on by (most cited first, 9 of 9)

- [Model Merging in the Essential Subspace](https://mlsanity.com/paper/2602.20208/) (2026)
- [FINE: Factorizing Knowledge for Initialization of Variable-sized Diffusion Models](https://mlsanity.com/paper/2409.19289/) (2024)
- [One-Shot Knowledge Transfer for Scalable Person Re-Identification](https://mlsanity.com/paper/2511.06016/) (2025)
- [Stratified Knowledge-Density Super-Network for Scalable Vision Transformers](https://mlsanity.com/paper/2511.11683/) (2025)
- [VT-Bench: A Unified Benchmark for Visual-Tabular Multi-Modal Learning](https://mlsanity.com/paper/2605.08146/) (2026)
- [Understanding Performance Collapse in Layer-Pruned Large Language Models via Decision Representation Transitions](https://mlsanity.com/paper/2605.07271/) (2026)
- [Dual-Model Weight Selection and Self-Knowledge Distillation for Medical Image Classification](https://mlsanity.com/paper/2508.20461/) (2025)
- [Constraint-based Pre-training: From Structured Constraints to Scalable Model Initialization](https://mlsanity.com/paper/2604.14769/) (2026)
- [Learngene Search Across Multiple Datasets for Building Variable-Sized Models](https://mlsanity.com/paper/2605.08209/) (2026)

## Abstract (arXiv)

In practice, we usually need to build variable-sized models adapting for diverse resource constraints in different application scenarios, where weight initialization is an important step prior to training. The Learngene framework, introduced recently, firstly learns one compact part termed as learngene from a large well-trained model, after which learngene is expanded to initialize variable-sized models. In this paper, we start from analysing the importance of guidance for the expansion of well-trained learngene layers, inspiring the design of a simple but highly effective Learngene approach termed SWS (Stage-wise Weight Sharing), where both learngene layers and their learning process critically contribute to providing knowledge and guidance for initializing models at varying scales. Specifically, to learn learngene layers, we build an auxiliary model comprising multiple stages where the layer weights in each stage are shared, after which we train it through distillation. Subsequently, we expand these learngene layers containing stage information at their corresponding stage to initialize models of variable depths. Extensive experiments on ImageNet-1K demonstrate that SWS achieves consistent better performance compared to many models trained from scratch, while reducing around 6.6x total training costs. In some cases, SWS performs better only after 1 epoch tuning. When initializing variable-sized models adapting for different resource constraints, SWS achieves better results while reducing around 20x parameters stored to initialize these models and around 10x pre-training costs, in contrast to the pre-training and fine-tuning approach.