MoECa: Aligning Feature Reuse with Expert Decomposition in Diffusion Transformers

• URL: http://arxiv.org/abs/2606.15615v1
• Date: Sun, 14 Jun 2026 05:54:36 GMT
• Title: MoECa: Aligning Feature Reuse with Expert Decomposition in Diffusion Transformers
• Abstract summary: MoECa is a fine-grained caching framework that performs branch-level feature reuse across timesteps.<n>MoECa consistently achieves a better speed-quality trade-off than prior caching methods, with up to 2.83$times$ inference speedup and minimal quality degradation.
• Score: 10.739058307716581
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Diffusion Transformers with Mixture-of-Experts (DiT-MoE) improve model capacity under sparse activation, but diffusion inference is still bottlenecked by redundant computation across timesteps. Existing caching methods mainly operate at the token level, which becomes suboptimal in DiT-MoE because each token update is internally decomposed into multiple routed expert branches. Our analysis shows that cross-timestep redundancy in DiT-MoE is better characterized at the expert-branch level than at the whole-token level. Based on this observation, we propose MoECa, a fine-grained caching framework that performs branch-level feature reuse across timesteps. MoECa further introduces expert-aware adaptive control and synchronized cache updates across MoE and attention paths to maintain stable intermediate states. Experiments on multiple DiT-MoE models show that MoECa consistently achieves a better speed-quality trade-off than prior caching methods, with up to 2.83$\times$ inference speedup and minimal quality degradation.

Related papers

• Tying the Loop -- Tied Expert Layers in Mixture-of-Experts Language Models [50.34585122347149]
  We introduce Expert Tying, an architectural modification that shares expert parameters across consecutive transformer layers.<n>We evaluate this approach across common, state-of-the-art architectures, including OLMoE, Qwen3, and DeepSeek-style MoEs.
  arXiv  Detail & Related papers  (2026-06-15T15:08:09Z)
• Sticky Routing: Training MoE Models for Memory-Efficient Inference [0.0]
  StickyMoE penalises abrupt expert switches between adjacent tokens.<n>Unlike post-hoc methods, StickyMoE allows expert representations and routing decisions to co-adapt from the first training step.
  arXiv  Detail & Related papers  (2026-06-12T23:47:19Z)
• Dynamic TMoE: A Drift-Aware Dynamic Mixture of Experts Framework for Non-Stationary Time Series Forecasting [39.362777320328256]
  We propose Dynamic TMoE, a framework that unifies architectural evolution with temporal continuity during learning phase.<n> Experiments on nine benchmarks demonstrate state-of-the-art performance, reducing MSE by 10.4% and MAE by 7.8%.
  arXiv  Detail & Related papers  (2026-05-20T03:45:22Z)
• ProCache: Constraint-Aware Feature Caching with Selective Computation for Diffusion Transformer Acceleration [14.306565517230775]
  Diffusion Transformers (DiTs) have achieved state-of-the-art performance in generative modeling, yet their high computational cost hinders real-time deployment.<n>Existing methods suffer from two key limitations: (1) uniform caching intervals fail to align with the non-uniform temporal dynamics of DiT, and (2) naive feature reuse with excessively large caching intervals can lead to severe error accumulation.<n>We propose ProCache, a training-free dynamic feature caching framework that addresses these issues via two core components.
  arXiv  Detail & Related papers  (2025-12-19T07:27:19Z)
• TNT: Improving Chunkwise Training for Test-Time Memorization [62.78875147721906]
  Recurrent neural networks (RNNs) with deep test-time memorization modules, such as Titans and TTT, represent a promising, linearly-scaling paradigm distinct from Transformers.<n>We introduce TNT, a novel training paradigm that decouples training efficiency from inference performance through a two-stage process.<n>TNT achieves a substantial acceleration in training speed-up to 17 times faster than the most accurate baseline configuration.
  arXiv  Detail & Related papers  (2025-11-10T17:45:09Z)
• Rewiring Experts on the Fly:Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert models [52.502867924372275]
  Mixture-of-Experts (MoE) models achieve efficient scaling through sparse expert activation, but often suffer from suboptimal routing decisions due to distribution shifts in deployment.<n>We propose textita data-free, online test-time framework that continuously adapts MoE routing decisions during text generation without external supervision or data.
  arXiv  Detail & Related papers  (2025-10-16T16:24:36Z)
• ResMoE: Space-efficient Compression of Mixture of Experts LLMs via Residual Restoration [61.579842548990754]
  Mixture-of-Experts (MoE) Transformer, the backbone of multiple phenomenal language models, leverages sparsity by activating only a fraction of model parameters for each input token.<n>We introduce ResMoE, an innovative MoE approximation framework that utilizes Wasserstein barycenter to extract a common expert (barycenter expert) and approximate the residuals between this barycenter expert and the original ones.
  arXiv  Detail & Related papers  (2025-03-10T03:15:54Z)
• M2R2: Mixture of Multi-Rate Residuals for Efficient Transformer Inference [8.792650582656913]
  We introduce Mixture of Multi-rate Residuals (M2R2), a framework that dynamically modulates residual velocity to improve early alignment.<n>M2R2 surpasses state-of-the-art distance-based strategies, balancing generation quality and speedup.<n>In self-speculative decoding setup, M2R2 achieves up to 2.8x speedups on MT-Bench.
  arXiv  Detail & Related papers  (2025-02-04T06:13:52Z)
• Towards Stabilized and Efficient Diffusion Transformers through Long-Skip-Connections with Spectral Constraints [51.83081671798784]
  Diffusion Transformers (DiT) have emerged as a powerful architecture for image and video generation, offering superior quality and scalability.<n>DiT's practical application suffers from inherent dynamic feature instability, leading to error amplification during cached inference.<n>We propose Skip-DiT, an image and video generative DiT variant enhanced with Long-Skip-Connections (LSCs) - the key efficiency component in U-Nets.
  arXiv  Detail & Related papers  (2024-11-26T17:28:10Z)
• Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching [56.286064975443026]
  We make an interesting and somehow surprising observation: the computation of a large proportion of layers in the diffusion transformer, through a caching mechanism, can be readily removed even without updating the model parameters. We introduce a novel scheme, named Learningto-Cache (L2C), that learns to conduct caching in a dynamic manner for diffusion transformers. Experimental results show that L2C largely outperforms samplers such as DDIM and DPM-r, alongside prior cache-based methods at the same inference speed.
  arXiv  Detail & Related papers  (2024-06-03T18:49:57Z)
• SMILE: Scaling Mixture-of-Experts with Efficient Bi-level Routing [47.11171833082974]
  We introduce SMILE, which exploits heterogeneous network bandwidth and splits a single-step routing into bi-level routing. Our experimental results show that the proposed method obtains a 2.5x speedup over Switch Transformer in terms of pretraining throughput on the Colossal Clean Crawled Corpus without losing any convergence speed.
  arXiv  Detail & Related papers  (2022-12-10T03:44:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.