paper-with-me

Papers

Dense Backpropagation Improves Training for Sparse Mixture-of-Experts

2025-04-16 · Ashwinee Panda, Vatsal Baherwani, Zain Sarwar, Benjamin Therien, Supriyo Chakraborty, Tom Goldstein

Mixture of Experts (MoE) pretraining is more scalable than dense Transformer pretraining, because MoEs learn to route inputs to a sparse set of their feedforward parameters. However, this means that MoEs only receive a sparse backward update, leading to training instability and suboptimal performance. We present a lightweight approximation method that gives the MoE router a dense gradient update while continuing to sparsely activate its parameters. Our method, which we refer to as Default MoE, substitutes missing expert activations with default outputs consisting of an exponential moving average of expert outputs previously seen over the course of training. This allows the router to receive signals from every expert for each token, leading to significant improvements in training performance. Our Default MoE outperforms standard TopK routing in a variety of settings without requiring significant computational overhead. Code: https://github.com/vatsal0/default-moe.

📄 PDF Abstract BibTeX arXiv:2504.12463

Code (1)

vatsal0/default-moe 공식 구현 pytorch

Tasks

Mixture-of-Experts

Methods 이 논문이 사용한 방법론

Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
Position-Wise Feed-Forward Layer 설명 없음
BPE Byte Pair Encoding, or BPE, is a subword segmentation algorithm that encodes rare and unknown words as sequences of subword units. The intuition is that various word…
Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Absolute Position Encodings Absolute Position Encodings are a type of position embeddings for [Transformer-based models] where positional encodings are…
Residual Connection 설명 없음
Label Smoothing Label Smoothing is a regularization technique that introduces noise for the labels. This accounts for the fact that datasets may have mistakes in them, so maximizing the…

Similar Papers 제목 키워드 기반

Sparse Backpropagation for MoE Training

2023-10-01 · Liyuan Liu, Jianfeng Gao, Weizhu Chen

One defining characteristic of Mixture-of-Expert (MoE) models is their capacity for conducting sparse computation via expert routing, leading to remarkable scalability. However, backpropagation, the cornerstone of deep l…

Machine Translation

GRIN: GRadient-INformed MoE

2024-09-18 · Liyuan Liu, Young Jin Kim, Shuohang Wang, Chen Liang 외

Mixture-of-Experts (MoE) models scale more effectively than dense models due to sparse computation through expert routing, selectively activating only a small subset of expert modules. However, sparse computation challen…

HellaSwagHumanEvalLanguage ModelingLanguage Modelling+3

CLIP-UP: A Simple and Efficient Mixture-of-Experts CLIP Training Recipe with Sparse Upcycling

2025-02-03 · Xinze Wang, Chen Chen, Yinfei Yang, Hong-You Chen 외

Mixture-of-Experts (MoE) models are crucial for scaling model capacity while controlling inference costs. While integrating MoE into multimodal models like CLIP improves performance, training these models is notoriously …

Mixture-of-Experts

Dense vs Sparse Pretraining at Tiny Scale: Active-Parameter vs Total-Parameter Matching

2026-05-13 · Abdalrahman Wael arxiv

We study dense and mixture-of-experts (MoE) transformers in a tiny-scale pretraining regime under a shared LLaMA-style decoder training recipe. The sparse model replaces dense feed-forward blocks with Mixtral-style route…

Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints

2022-12-09 · Aran Komatsuzaki, Joan Puigcerver, James Lee-Thorp, Carlos Riquelme Ruiz 외

Training large, deep neural networks to convergence can be prohibitively expensive. As a result, often only a small selection of popular, dense models are reused across different contexts and tasks. Increasingly, sparsel…

Mixture-of-Experts