# Upcycling Large Language Models into Mixture of Experts

arXiv: https://arxiv.org/abs/2410.07524 · HTML: https://mlsanity.com/paper/2410.07524/

## Key facts

- Cited by 49 papers in the map (data through 14 Oct 2026).
- Placed in MoE large language models, one of 12 directions in Mixture-of-experts.
- Builds on 39 of its 41 references that are in the map.
- First posted on arXiv 10 Oct 2024.

## Placed in

- Mixture-of-experts / [MoE large language models](https://mlsanity.com/moe/moe-large-language-models/)

## Builds on (most cited first, 30 of 39)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Qwen Technical Report](https://mlsanity.com/paper/2309.16609/) (2023)
- [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://mlsanity.com/paper/2101.03961/) (2021)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://mlsanity.com/paper/1909.08053/) (2019)
- [Mixtral of Experts](https://mlsanity.com/paper/2401.04088/) (2024)
- [GLU Variants Improve Transformer](https://mlsanity.com/paper/2002.05202/) (2020)
- [GShard: Scaling Giant Models with Conditional Computation and Automatic Sharding](https://mlsanity.com/paper/2006.16668/) (2020)
- [Scaling Language Models: Methods, Analysis & Insights from Training Gopher](https://mlsanity.com/paper/2112.11446/) (2021)
- [DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model](https://mlsanity.com/paper/2405.04434/) (2024)
- [GLaM: Efficient Scaling of Language Models with Mixture-of-Experts](https://mlsanity.com/paper/2112.06905/) (2021)
- [DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models](https://mlsanity.com/paper/2401.06066/) (2024)
- [DeepSeek LLM: Scaling Open-Source Language Models with Longtermism](https://mlsanity.com/paper/2401.02954/) (2024)
- [Mixture-of-Experts with Expert Choice Routing](https://mlsanity.com/paper/2202.09368/) (2022)
- [ST-MoE: Designing Stable and Transferable Sparse Expert Models](https://mlsanity.com/paper/2202.08906/) (2022)
- [OLMoE: Open Mixture-of-Experts Language Models](https://mlsanity.com/paper/2409.02060/) (2024)
- [NeMo: a toolkit for building AI applications using Neural Modules](https://mlsanity.com/paper/1909.09577/) (2019)
- [MegaBlocks: Efficient Sparse Training with Mixture-of-Experts](https://mlsanity.com/paper/2211.15841/) (2022)
- [Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints](https://mlsanity.com/paper/2212.05055/) (2022)
- [OpenMoE: An Early Effort on Open Mixture-of-Experts Language Models](https://mlsanity.com/paper/2402.01739/) (2024)
- [Scaling Laws for Fine-Grained Mixture of Experts](https://mlsanity.com/paper/2402.07871/) (2024)
- [On the Representation Collapse of Sparse Mixture of Experts](https://mlsanity.com/paper/2204.09179/) (2022)
- [A Review of Sparse Expert Models in Deep Learning](https://mlsanity.com/paper/2209.01667/) (2022)
- [Nemotron-4 340B Technical Report](https://mlsanity.com/paper/2406.11704/) (2024)
- [Branch-Train-MiX: Mixing Expert LLMs into a Mixture-of-Experts LLM](https://mlsanity.com/paper/2403.07816/) (2024)
- [LLaMA Pro: Progressive LLaMA with Block Expansion](https://mlsanity.com/paper/2401.02415/) (2024)
- [Learning to Grow Pretrained Models for Efficient Transformer Training](https://mlsanity.com/paper/2303.00980/) (2023)
- [Reuse, Don't Retrain: A Recipe for Continued Pretraining of Language Models](https://mlsanity.com/paper/2407.07263/) (2024)

## Built on by (most cited first, 30 of 49)

- [Step-3 is Large yet Affordable: Model-system Co-design for Cost-effective Decoding](https://mlsanity.com/paper/2507.19427/) (2025)
- [LLaMA-MoE v2: Exploring Sparsity of LLaMA from Perspective of Mixture-of-Experts with Post-Training](https://mlsanity.com/paper/2411.15708/) (2024)
- [Drop-Upcycling: Training Sparse Mixture of Experts with Partial Re-initialization](https://mlsanity.com/paper/2502.19261/) (2025)
- [Make LoRA Great Again: Boosting LoRA with Adaptive Singular Values and Mixture-of-Experts Optimization Alignment](https://mlsanity.com/paper/2502.16894/) (2025)
- [Scalable Training of Mixture-of-Experts Models with Megatron Core](https://mlsanity.com/paper/2603.07685/) (2026)
- [MoE Parallel Folding: Heterogeneous Parallelism Mappings for Efficient Large-Scale MoE Model Training with Megatron Core](https://mlsanity.com/paper/2504.14960/) (2025)
- [Scaling Laws for Upcycling Mixture-of-Experts Language Models](https://mlsanity.com/paper/2502.03009/) (2025)
- [DeRS: Towards Extremely Efficient Upcycled Mixture-of-Experts Models](https://mlsanity.com/paper/2503.01359/) (2025)
- [Mixture-of-Experts Can Surpass Dense LLMs Under Strictly Equal Resource](https://mlsanity.com/paper/2506.12119/) (2025)
- [FreqMoE: Dynamic Frequency Enhancement for Neural PDE Solvers](https://mlsanity.com/paper/2505.06858/) (2025)
- [LIBMoE: A Library for comprehensive benchmarking Mixture of Experts in Large Language Models](https://mlsanity.com/paper/2411.00918/) (2024)
- [MoORE: SVD-based Model MoE-ization for Conflict- and Oblivion-Resistant Multi-Task Adaptation](https://mlsanity.com/paper/2506.14436/) (2025)
- [SpikingBrain: Spiking Brain-inspired Large Models](https://mlsanity.com/paper/2509.05276/) (2025)
- [X-EcoMLA: Upcycling Pre-Trained Attention into MLA for Efficient and Extreme KV Compression](https://mlsanity.com/paper/2503.11132/) (2025)
- [Omni-Router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition](https://mlsanity.com/paper/2507.05724/) (2025)
- [From Curated Data to Scalable Models: Continual Pre-training of Dense and MoE Large Language Models for Tibetan](https://mlsanity.com/paper/2507.09205/) (2025)
- [Innovator: Scientific Continued Pretraining with Fine-grained MoE Upcycling](https://mlsanity.com/paper/2507.18671/) (2025)
- [MoME: Mixture of Matryoshka Experts for Audio-Visual Speech Recognition](https://mlsanity.com/paper/2510.04136/) (2025)
- [Sparse Checkpointing for Fast and Reliable MoE Training](https://mlsanity.com/paper/2412.15411/) (2024)
- [The Race to Efficiency: A New Perspective on AI Scaling Laws](https://mlsanity.com/paper/2501.02156/) (2025)
- [Mix-of-Language-Experts Architecture for Multilingual Programming](https://mlsanity.com/paper/2506.18923/) (2025)
- [AWM: Accurate Weight-Matrix Fingerprint for Large Language Models](https://mlsanity.com/paper/2510.06738/) (2025)
- [Quant Experts: Token-aware Adaptive Error Reconstruction with Mixture of Experts for Large Vision-Language Models Quantization](https://mlsanity.com/paper/2602.24059/) (2026)
- [Marco-MoE: Open Multilingual Mixture-of-Expert Language Models with Efficient Upcycling](https://mlsanity.com/paper/2604.25578/) (2026)
- [Scaling Fine-Grained MoE Beyond 50B Parameters: Empirical Evaluation and Practical Insights](https://mlsanity.com/paper/2506.02890/) (2025)
- [Dirichlet-Prior Shaping: Guiding Expert Specialization in Upcycled MoEs](https://mlsanity.com/paper/2510.01185/) (2025)
- [Matrix-Driven Identification and Reconstruction of LLM Weight Homology](https://mlsanity.com/paper/2508.06309/) (2025)
- [Router Upcycling: Leveraging Mixture-of-Routers in Mixture-of-Experts Upcycling](https://mlsanity.com/paper/2509.00679/) (2025)
- [Symphony-MoE: Harmonizing Disparate Pre-trained Models into a Coherent Mixture-of-Experts](https://mlsanity.com/paper/2509.18542/) (2025)
- [Beyond Sunk Costs: Boosting LLM Pre-training Efficiency via Orthogonal Growth of Mixture-of-Experts](https://mlsanity.com/paper/2510.08008/) (2025)

## Abstract (arXiv)

Upcycling pre-trained dense language models into sparse mixture-of-experts (MoE) models is an efficient approach to increase the model capacity of already trained models. However, optimal techniques for upcycling at scale remain unclear. In this work, we conduct an extensive study of upcycling methods and hyperparameters for billion-parameter scale language models. We propose a novel "virtual group" initialization scheme and weight scaling approach to enable upcycling into fine-grained MoE architectures. Through ablations, we find that upcycling outperforms continued dense model training. In addition, we show that softmax-then-topK expert routing improves over topK-then-softmax approach and higher granularity MoEs can help improve accuracy. Finally, we upcycled Nemotron-4 15B on 1T tokens and compared it to a continuously trained version of the same model on the same 1T tokens: the continuous trained model achieved 65.3% MMLU, whereas the upcycled model achieved 67.6%. Our results offer insights and best practices to effectively leverage upcycling for building MoE language models. Code is available.