# Dynamic Expert Sharing: Decoupling Memory from Parallelism in Mixture-of-Experts Diffusion LLMs

arXiv: https://arxiv.org/abs/2602.00879 · HTML: https://mlsanity.com/paper/2602.00879/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Efficient MoE training and inference systems, one of 12 directions in Mixture-of-experts.
- Builds on 31 of its 34 references that are in the map.
- First posted on arXiv 31 Jan 2026.

## Placed in

- Mixture-of-experts / [Efficient MoE training and inference systems](https://mlsanity.com/moe/efficient-moe-training-and-inference-systems/)
- Efficient LLM inference and serving / [Diffusion LLM and MoE inference](https://mlsanity.com/inference/diffusion-llm-and-moe-inference/)

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

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [Program Synthesis with Large Language Models](https://mlsanity.com/paper/2108.07732/) (2021)
- [gpt-oss-120b & gpt-oss-20b Model Card](https://mlsanity.com/paper/2508.10925/) (2025)
- [Dream 7B: Diffusion Large Language Models](https://mlsanity.com/paper/2508.15487/) (2025)
- [Non-Autoregressive Neural Machine Translation](https://mlsanity.com/paper/1711.02281/) (2017)
- [Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding](https://mlsanity.com/paper/2505.22618/) (2025)
- [Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models](https://mlsanity.com/paper/2503.09573/) (2025)
- [Break the Sequential Dependency of LLM Inference Using Lookahead Decoding](https://mlsanity.com/paper/2402.02057/) (2024)
- [LLaDA2.0: Scaling Up Diffusion Language Models to 100B](https://mlsanity.com/paper/2512.15745/) (2025)
- [Not All Experts are Equal: Efficient Expert Pruning and Skipping for Mixture-of-Experts Large Language Models](https://mlsanity.com/paper/2402.14800/) (2024)
- [SDAR: A Synergistic Diffusion-AutoRegression Paradigm for Scalable Sequence Generation](https://mlsanity.com/paper/2510.06303/) (2025)
- [Accelerating Transformer Inference for Translation via Parallel Decoding](https://mlsanity.com/paper/2305.10427/) (2023)
- [Merge, Then Compress: Demystify Efficient SMoE with Hints from Its Routing Policy](https://mlsanity.com/paper/2310.01334/) (2023)
- [MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs](https://mlsanity.com/paper/2411.11217/) (2024)
- [Task-Specific Expert Pruning for Sparse Mixture-of-Experts](https://mlsanity.com/paper/2206.00277/) (2022)
- [LLaDA-MoE: A Sparse MoE Diffusion Language Model](https://mlsanity.com/paper/2509.24389/) (2025)
- [Mixture Compressor for Mixture-of-Experts LLMs Gains More](https://mlsanity.com/paper/2410.06270/) (2024)
- [Diversifying the Expert Knowledge for Task-Agnostic Pruning in Sparse Mixture-of-Experts](https://mlsanity.com/paper/2407.09590/) (2024)
- [dInfer: An Efficient Inference Framework for Diffusion Language Models](https://mlsanity.com/paper/2510.08666/) (2025)
- [AdaBlock-dLLM: Semantic-Aware Diffusion LLM Inference via Adaptive Block Size](https://mlsanity.com/paper/2509.26432/) (2025)
- [Set Block Decoding is a Language Model Inference Accelerator](https://mlsanity.com/paper/2509.04185/) (2025)
- [Hardware-Aware Parallel Prompt Decoding for Memory-Efficient Acceleration of LLM Inference](https://mlsanity.com/paper/2405.18628/) (2024)
- [DA-MoE: Towards Dynamic Expert Allocation for Mixture-of-Experts Models](https://mlsanity.com/paper/2409.06669/) (2024)
- [BiTA: Bi-Directional Tuning for Lossless Acceleration in Large Language Models](https://mlsanity.com/paper/2401.12522/) (2024)
- [MoDES: Accelerating Mixture-of-Experts Multimodal Large Language Models via Dynamic Expert Skipping](https://mlsanity.com/paper/2511.15690/) (2025)
- [Challenges and Research Directions for Large Language Model Inference Hardware](https://mlsanity.com/paper/2601.05047/) (2026)
- [Opportunistic Expert Activation: Batch-Aware Expert Routing for Faster Decode Without Retraining](https://mlsanity.com/paper/2511.02237/) (2025)

## Built on by (most cited first, 4 of 4)

- [dMoE: dLLMs with Learnable Block Experts](https://mlsanity.com/paper/2605.30876/) (2026)
- [AcceptMoE: Commitment-Weighted Self-Sizing Verifier Expert Sets for Efficient MoE Speculative Decoding](https://mlsanity.com/paper/2608.02989/) (2026)
- [OLED-MoE: Accelerating MoE-Based dLLM Inference via Inter-Iteration Locality-Aware Expert Offloading](https://mlsanity.com/paper/2609.33385/) (2026)
- [BASE: Batch-Aware Selection of Experts Using Predicted Removal Error for Efficient MoE Decoding](https://mlsanity.com/paper/2609.36222/) (2026)

## Abstract (arXiv)

Among parallel decoding paradigms, diffusion large language models (dLLMs) have emerged as a promising candidate that balances generation quality and throughput. However, their integration with Mixture-of-Experts (MoE) architectures is constrained by an expert explosion: as the number of tokens generated in parallel increases, the number of distinct experts activated grows nearly linearly. This results in substantial memory traffic that pushes inference into a memory-bound regime, negating the efficiency gains of both MoE and parallel decoding. To address this challenge, we propose Dynamic Expert Sharing (DES), a novel technique that shifts MoE optimization from token-centric pruning and conventional expert skipping methods to sequence-level coreset selection. To maximize expert reuse, DES identifies a compact, high-utility set of experts to satisfy the requirements of an entire parallel decoding block. We introduce two innovative selection strategies: (1) Intra-Sequence Sharing (DES-Seq), which adapts optimal allocation to the sequence level, and (2) Saliency-Aware Voting (DES-Vote), a novel mechanism that allows tokens to collectively elect a coreset based on aggregated router weights. Extensive experiments on MoE dLLMs demonstrate that DES reduces unique expert activations by over 55% and latency by up to 38%, while retaining 99% of vanilla accuracy, effectively decoupling memory overhead from the degree of parallelism.