paper-with-me

홈 › Papers

Pruning Neural Machine Translation for Speed Using Group Lasso

2021-11-01 · WMT (EMNLP) 2021 11 · Maximiliana Behnke, Kenneth Heafield

Unlike most work on pruning neural networks, we make inference faster. Group lasso regularisation enables pruning entire rows, columns or blocks of parameters that result in a smaller dense network. Because the network is still dense, efficient matrix multiply routines are still used and only minimal software changes are required to support variable layer sizes. Moreover, pruning is applied during training so there is no separate pruning step. Experiments on top of English->German models, which already have state-of-the-art speed and size, show that two-thirds of feedforward connections can be removed with 0.2 BLEU loss. With 6 decoder layers, the pruned model is 34% faster; with 2 tied decoder layers, the pruned model is 14% faster. Pruning entire heads and feedforward connections in a 12–1 encoder-decoder architecture gains an additional 51% speed-up. These push the Pareto frontier with respect to the trade-off between time and quality compared to strong baselines. In the WMT 2021 Efficiency Task, our pruned and quantised models are 1.9–2.7x faster at the cost 0.9–1.7 BLEU in comparison to the unoptimised baselines. Across language pairs, we see similar sparsity patterns: an ascending or U-shaped distribution in encoder feedforward and attention layers and an ascending distribution in the decoder.

📄 PDF Abstract BibTeX

Code (0)

등록된 구현이 없습니다.

Tasks

DecoderMachine TranslationTranslation

Methods 이 논문이 사용한 방법론

Pruning 설명 없음

Similar Papers 제목 키워드 기반

Directed-Weighting Group Lasso for Eltwise Blocked CNN Pruning

2019-10-21 · Ke Zhan, Shimiao Jiang, Yu Bai, Yi Li 외

Eltwise layer is a commonly used structure in the multi-branch deep learning network. In a filter-wise pruning procedure, due to the specific operation of the eltwise layer, all its previous convolutional layers should v…

Smooth Bilevel Programming for Sparse Regularization

2021-06-02 · NeurIPS 2021 12 · Clarice Poon, Gabriel Peyré

Iteratively reweighted least square (IRLS) is a popular approach to solve sparsity-enforcing regression problems in machine learning. State of the art approaches are more efficient but typically rely on specific coordina…

Specificity

Speeding Up Neural Machine Translation Decoding by Cube Pruning

2018-09-09 · EMNLP 2018 10 · Wen Zhang, Liang Huang, Yang Feng, Lei Shen 외

Although neural machine translation has achieved promising results, it suffers from slow translation speed. The direct consequence is that a trade-off has to be made between translation quality and speed, thus its perfor…

Machine TranslationTranslation

SIMD-size aware weight regularization for fast neural vocoding on CPU

2022-11-02 · Hiroki Kanagawa, Yusuke Ijima

This paper proposes weight regularization for a faster neural vocoder. Pruning time-consuming DNN modules is a promising way to realize a real-time vocoder on a CPU (e.g. WaveRNN, LPCNet). Regularization that encourages …

CPU

Adaptive Optimizers with Sparse Group Lasso for Neural Networks in CTR Prediction

2021-07-30 · Yun Yue, Yongchao Liu, Suo Tong, Minghao Li 외

We develop a novel framework that adds the regularizers of the sparse group lasso to a family of adaptive optimizers in deep learning, such as Momentum, Adagrad, Adam, AMSGrad, AdaHessian, and create a new class of optim…

Click-Through Rate Prediction