REVIEW 8 cited by
Soft Merging of Experts with Adaptive Routing
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Sparsely activated neural networks with conditional computation learn to route their inputs through different "expert" subnetworks, providing a form of modularity that densely activated models lack. Despite their possible benefits, models with learned routing often underperform their parameter-matched densely activated counterparts as well as models that use non-learned heuristic routing strategies. In this paper, we hypothesize that these shortcomings stem from the gradient estimation techniques used to train sparsely activated models that use non-differentiable discrete routing decisions. To address this issue, we introduce Soft Merging of Experts with Adaptive Routing (SMEAR), which avoids discrete routing by using a single "merged" expert constructed via a weighted average of all of the experts' parameters. By routing activations through a single merged expert, SMEAR does not incur a significant increase in computational costs and enables standard gradient-based training. We empirically validate that models using SMEAR outperform models that route based on metadata or learn sparse routing through gradient estimation. Furthermore, we provide qualitative analysis demonstrating that the experts learned via SMEAR exhibit a significant amount of specialization. All of the code used in our experiments is publicly available.
Forward citations
Cited by 8 Pith papers
-
It Takes a MAESTRO To Prune Bad Experts
Pruning MoE LLMs according to the stationary distribution of a Markov chain over (layer, expert) routing transitions retains more task performance than local importance heuristics, with up to ~3.5% relative gains over...
-
CoMoL: Efficient Mixture of LoRA Experts via Dynamic Core Space Merging
CoMoL represents every LoRA expert as a shared-basis core matrix and merges token-selected experts in that core space, reaching standard LoRA parameter counts while outperforming MoE-LoRA baselines on math and code.
-
BlockFFN: Towards End-Side Acceleration-Friendly Mixture-of-Experts with Chunk-Level Activation Sparsity
A ReLU-routed MoE with chunk-level sparsity training objectives and custom kernels combining activation sparsity with speculative decoding achieves over 70% 8-token chunk sparsity and up to 3.67x end-side speedup.
-
Automatic Expert Discovery in LLM Upcycling via Sparse Interpolated Mixture-of-Experts
Sparse Interpolated Mixture-of-Experts learns neuron-level masks and a per-prompt router to upcycle a dense LLM into a sparse expert model, improving instruction-tuning benchmarks with lower inference cost.
-
MetricHMSR:Metric Human Mesh and Scene Recovery from Monocular Images
From one image, MetricHMSR jointly estimates a metric human mesh, its global 3D position, and a corrected metric scene depth map.
-
Guiding the Experts: Semantic Priors for Efficient and Focused MoE Routing
Guiding Soft MoE dispatch weights with foreground segmentation masks plus a zero-initialized LayerScale improves ImageNet-1K top-1 by 0.6% and ImageNet-100 by 1.4% over a reproduced baseline.
-
Modular Foundation Models for Time-Series Perception in Digital Twins
A gated bank of frozen self-supervised time-series encoders, aligned and aggregated by a Transformer, supports competitive multi-task perception for digital twins and hydro-generator virtual sensing.
-
Stay Unique, Stay Efficient: Preserving Model Personality in Multi-Task Merging
DTS approximates task-specific model deltas with truncated SVD plus grouped thresholding and scaling, achieving near-individual multi-task accuracy with ~1% extra storage per task.
Discussion (0). Sign in to comment.