paper-with-me

Papers

Efficient softmax approximation for GPUs

2016-09-14 · ICML 2017 8 · Edouard Grave, Armand Joulin, Moustapha Cissé, David Grangier, Hervé Jégou

We propose an approximate strategy to efficiently train neural network based language models over very large vocabularies. Our approach, called adaptive softmax, circumvents the linear dependency on the vocabulary size by exploiting the unbalanced word distribution to form clusters that explicitly minimize the expectation of computation time. Our approach further reduces the computational time by exploiting the specificities of modern architectures and matrix-matrix vector operations, making it particularly suited for graphical processing units. Our experiments carried out on standard benchmarks, such as EuroParl and One Billion Word, show that our approach brings a large gain in efficiency over standard approximations while achieving an accuracy close to that of the full softmax. The code of our method is available at https://github.com/facebookresearch/adaptive-softmax.

📄 PDF Abstract BibTeX arXiv:1609.04309

Code (12)

facebookresearch/adaptive-softmax 공식 구현 torch
DavidWBressler/adaptivesoftmax pytorch
Jmkernes/PAR-Transformer-XL tf
ahmedbahaaeldin/Papers-from-Scratch
astanway/gated-conv-nets tf
cedrickchee/pytorch-pretrained-BERT pytorch
huggingface/transformers pytorch
jiali-ms/JLM tf
rdspring1/PyTorch_GBW_LM pytorch
rosinality/adaptive-softmax-pytorch pytorch
simon555/LM_word pytorch
yangsaiyong/tf-adaptive-softmax-lstm-lm tf

Methods 이 논문이 사용한 방법론

Adaptive Softmax Adaptive Softmax is a speedup technique for the computation of probability distributions over words. The adaptive softmax is…

Similar Papers 제목 키워드 기반

FLASH-D: FlashAttention with Hidden Softmax Division

2025-05-20 · Kosmas Alexandridis, Vasileios Titopoulos, Giorgos Dimitrakopoulos

The transformer's attention mechanism has revolutionized AI and machine learning, with its efficient computation being crucial to its performance. However, calculating attention involves matrix operations interspersed wi…

MonarchAttention: Zero-Shot Conversion to Fast, Hardware-Aware Structured Attention

2025-05-24 · Can Yaras, Alec S. Xu, Pierre Abillama, Changwoo Lee 외

Transformers have achieved state-of-the-art performance across various tasks, but suffer from a notable quadratic complexity in sequence length due to the attention mechanism. In this work, we propose MonarchAttention --…

16k4k

SoftmAP: Software-Hardware Co-design for Integer-Only Softmax on Associative Processors

2024-11-26 · Mariam Rakka, Jinhao Li, Guohao Dai, Ahmed Eltawil 외

Recent research efforts focus on reducing the computational and memory overheads of Large Language Models (LLMs) to make them feasible on resource-constrained devices. Despite advancements in compression techniques, non-…

QuantizationSensitivity

Revisiting Softmax for Uncertainty Approximation in Text Classification

2021-11-16 · ACL ARR November 2021 11 · Anonymous

Uncertainty approximation in text classification is an important area with applications in domain adaptation and interpretability. The most widely used uncertainty approximation method is Monte Carlo Dropout, which is co…

ClassificationDomain Adaptationtext-classificationText Classification

Revisiting Softmax for Uncertainty Approximation in Text Classification

2022-10-25 · Andreas Nugaard Holm, Dustin Wright, Isabelle Augenstein

Uncertainty approximation in text classification is an important area with applications in domain adaptation and interpretability. One of the most widely used uncertainty approximation methods is Monte Carlo (MC) Dropout…

ClassificationDomain Adaptationtext-classificationText Classification