Challenge: Nevertheless, training from scratch on trillions of tokens remains expensive that most users can only finetune these models.
Approach: They propose to reuse parameters of dense models for the MoE layers with a router that can integrate new experts into existing trained models without hurting performance on previous domains.
Outcome: The proposed router can integrate new experts into existing trained models without hurting the performance on previous domains.

Similar Papers

Upcycling Instruction Tuning from Dense to Mixture-of-Experts via Parameter Merging (2025.acl-long)

Copied to clipboard

Challenge: Existing methods for tuning large language models from dense to MoE face significant data requirements and require large-scale post-training.
Approach: They propose an upcycling instruction tuning approach for tuning a dense pre-trained model into a MoE instruction model using genetic algorithm and parameter merging.
Outcome: The proposed approach improves the performance of large language models with a small amount of seed data and improves their scaling.
A Closer Look into Mixture-of-Experts in Large Language Models (2025.findings-naacl)

Copied to clipboard

Challenge: Mixture-of-experts (MoE) architectures are gaining increasing attention for their unique properties and remarkable performance.
Approach: They propose a mixture-of-experts architecture that allows for model scaling without sacrificing computational efficiency.
Outcome: The proposed model increases model size without sacrificing computational efficiency . the proposed model is modular and can be used by a broad spectrum of practitioners .
Improved Sparse Upcycling for Instruction Tuning (2025.coling-main)

Copied to clipboard

Challenge: Existing methods for sparse upcycling lead to performance degradation in instruction tuning scenarios.
Approach: They propose a representation-based approach to convert dense language models into sparsely activated ones by initializing router weights from language models.
Outcome: The proposed architecture improves model capabilities and routing consistency across multiple benchmarks.
HyperRouter: Towards Efficient Training and Inference of Sparse Mixture of Experts (2023.emnlp-main)

Copied to clipboard

Challenge: Recent studies suggest that fixing the routers can achieve competitive performance by alleviating the collapsing problem, where all experts eventually learn similar representations.
Approach: They propose a method that dynamically generates router parameters through a fixed hypernetwork and trainable embeddings to achieve a balance between training the routers and freezing them to learn an improved routing policy.
Outcome: Experiments on a wide range of tasks show that the proposed method performs better than existing methods.
Parameter-Efficient Routed Fine-Tuning: Mixture-of-Experts Demands Mixture of Adaptation Modules (2026.findings-eacl)

Copied to clipboard

Challenge: Existing Parameter-Efficient Fine-Tuning (PEFT) strategies that focus on specialized experts are not effective for Mixture-of-Experts (MoE).
Approach: They propose to integrate a dynamic routing mechanism among specialized experts in Mixture-of-Experts (MoE) .
Outcome: Extensive experiments on commonsense and math reasoning tasks validate the performance and efficiency of the proposed routed approach.
LoRACoE: Improving Large Language Model via Composition-based LoRA Expert (2025.emnlp-main)

Copied to clipboard

Challenge: Recent studies show that the Mixture of Experts architecture improves performance of large language models.
Approach: They propose a method to build static experts using LoRA parameters . they propose to use rank-level parameters to build experts based on rank-based parameters based in LoRA module.
Outcome: The proposed method improves task performance across a broader range of tasks.
Union-of-Experts: Neurons in Mixture-of-Experts are Secretly Routers (2026.acl-long)

Copied to clipboard

Challenge: Mixture-of-Experts (MoE) models rely on an external router to assign tokens to experts, resulting in suboptimal performance.
Approach: They propose an MoE variant that performs "expert-autonomous routing" by pre-designating a fraction of neurons within each expert as "routing neurons" they pre-train UoE models with up to 3B parameters and show they outperform traditional MoEs with matched efficiency.
Outcome: The proposed model outperforms existing models with 3B parameters and provides valuable insights into expert-autonomous selection and the broader routing mechanisms of MoE models.
MixtureKit: A General Framework for Composing, Training, and Visualizing Mixture-of-Experts Models (2026.acl-demo)

Copied to clipboard

Challenge: MixtureKit is a modular open-source framework for constructing, training, and analyzing Mixture-of-Experts (MoE) models from arbitrary pre-trained or fine-tuned checkpoints.
Approach: They propose a modular open-source framework for constructing, training, and analyzing Mixture-of-Experts (MoE) models from arbitrary pre-trained or fine-tuned checkpoints.
Outcome: Experiments on multilingual code-switched (Arabic–Latin) show that BTX models built with MixtureKit outperform dense baselines across multiple benchmarks.
Mixture of Diverse Size Experts (2024.emnlp-industry)

Copied to clipboard

Challenge: Recent large language models (LLMs) have shown superior performance in a variety of tasks due to the sub-linearly increasing computational costs.
Approach: They propose a new MoE architecture with designed layers where experts have different sizes to mitigate this defect.
Outcome: The proposed architecture surpasses existing MoEs by adaptively assigning the parameter budget to experts while maintaining the same total parameter size and number of experts.
SAMoRA: Semantic-Aware Mixture of LoRA Experts for Task-Adaptive Learning (2026.findings-acl)

Copied to clipboard

Challenge: Existing methods for multitask learning fail to match input semantics with expert capabilities, leading to weak expert specialization.
Approach: They propose a parameter-efficient mixture-of-experts framework for task-adaptive learning that aligns textual semantics with the most suitable experts for precise routing.
Outcome: The proposed framework outperforms the state-of-the-art methods and holds excellent task generalization capabilities.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations