AI Paper Atlas

Research map / Architectures and efficiency / Mixture of experts

TD-MoE: Tensor Decomposition for MoE Models

Yuebin XU, YANHONG WANG, Xuemei Peng, Hui Zang, Chen Minghao, Pengfei Xia, Zeyi Wen

ICLR 2026 · Architectures and efficiency › Mixture of experts · approach: Expert routing and specialization

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace per-expert SVD with joint Tucker decomposition (output whitening)MoE-SVD, Qwen2-57B-A14B · 40% compression avg accuracyavg accuracy: 0.48 -> 0.56Improved
replace per-expert SVD with joint Tucker decomposition (output whitening)MoE-SVD, Mixtral-8x7B · 40% compression avg accuracyavg accuracy: 0.50 -> 0.57Improved
replace per-expert SVD with joint Tucker decomposition (input whitening)MoE-SVD, Mixtral-8x7B · 60% compression avg accuracyavg accuracy: 0.37 -> 0.45Improved
replace per-expert SVD with joint Tucker decomposition (input whitening)MoE-SVD, Qwen2-57B-A14B · 60% compression avg accuracyavg accuracy: 0.47 -> 0.52Improved
replace per-expert SVD with joint Tucker decomposition (output whitening)MoE-SVD, Mixtral-8x7B · 20% compression WikiText-2perplexity: 4.86 -> 4.49Improved
use input whitening instead of output whitening (ablation)TD-MoE, Mixtral-8x7B · 60% compression avg accuracyavg accuracy: 0.42 -> 0.45Improved
use output whitening instead of input whitening (ablation)TD-MoE, Mixtral-8x7B · 40% compression avg accuracyavg accuracy: 0.54 -> 0.57Improved
add NF4 quantization after decompositionTD-MoE, Mixtral-8x7B · 20% compressionLM avg perplexity: 9.81 -> 9.70No clear effect
add 20% structured pruning of Tucker coreTD-MoE, Mixtral-8x7B · 20% compressionreasoning avg accuracy: 0.66 -> 0.64Hurt
add 40% structured pruning of Tucker coreTD-MoE, Mixtral-8x7B · 20% compressionreasoning avg accuracy: 0.66 -> 0.61Hurt
add multi-linear whiteningTD-MoE · covariance conditioningpost-whitening deviation: 9 - 5.4x10^4 -> 1.1x10^-7Improved
vary calibration set size N from 128 to 2k (ablation)TD-MoE · 20% compressionavg PPL: 9.81-9.84No clear effect

Later papers that compared with this method

1 later accepted paper reports TD-MoE as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy