paper-with-me

Papers

Exploring Extreme Parameter Compression for Pre-trained Language Models

2022-05-20 · ICLR 2022 4 · Yuxin Ren, Benyou Wang, Lifeng Shang, Xin Jiang, Qun Liu

Recent work explored the potential of large-scale Transformer-based pre-trained models, especially Pre-trained Language Models (PLMs) in natural language processing. This raises many concerns from various perspectives, e.g., financial costs and carbon emissions. Compressing PLMs like BERT with negligible performance loss for faster inference and cheaper deployment has attracted much attention. In this work, we aim to explore larger compression ratios for PLMs, among which tensor decomposition is a potential but under-investigated one. Two decomposition and reconstruction protocols are further proposed to improve the effectiveness and efficiency during compression. Our compressed BERT with ${1}/{7}$ parameters in Transformer layers performs on-par with, sometimes slightly better than the original BERT in GLUE benchmark. A tiny version achieves $96.7\%$ performance of BERT-base with $ {1}/{48} $ encoder parameters (i.e., less than 2M parameters excluding the embedding layer) and $2.7 \times$ faster on inference. To show that the proposed method is orthogonal to existing compression methods like knowledge distillation, we also explore the benefit of the proposed method on a distilled BERT.

📄 PDF Abstract BibTeX arXiv:2205.10036

Code (1)

twinkle0331/xcompression 공식 구현 pytorch

Tasks

Knowledge DistillationTensor Decomposition

Methods 이 논문이 사용한 방법론

Multi-Head Attention 설명 없음
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
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…
Weight Decay 설명 없음
Dense Connections Dense Connections, or Fully Connected Connections, are a type of layer in a deep neural network that use a linear operation where every input is connected to every output…
Linear Warmup With Linear Decay Linear Warmup With Linear Decay is a learning rate schedule in which we increase the learning rate linearly for $n$ updates and then linearly decay afterwards.
Dropout Dropout is a regularization technique for neural networks that drops a unit (along with connections) at training time with a specified probability $p$ (a common value is…
Refunds@Expedia|||How do I get a full refund from Expedia? “How do I get a full refund from Expedia? How do I get a full refund from Expedia? – Call ☎️ +1-(888) 829 (0881) or +1-805-330-4056 or +1-805-330-4056 for Quick Help &…

Similar Papers 제목 키워드 기반

What Happens When Small Is Made Smaller? Exploring the Impact of Compression on Small Data Pretrained Language Models

2024-04-06 · Busayo Awobade, Mardiyyah Oduwole, Steven Kolawole

Compression techniques have been crucial in advancing machine learning by enabling efficient training and deployment of large-scale language models. However, these techniques have received limited attention in the contex…

Knowledge DistillationLanguage ModelingLanguage ModellingQuantization

Dynamic Base model Shift for Delta Compression

2025-05-16 · Chenyu Huang, Peng Ye, Shenghe Zheng, Xiaohui Wang 외

Transformer-based models with the pretrain-finetune paradigm bring about significant progress, along with the heavy storage and deployment costs of finetuned models on multiple tasks. Delta compression attempts to lower …

model

LLMC+: Benchmarking Vision-Language Model Compression with a Plug-and-play Toolkit

2025-08-13 · Chengtao Lv, Bilang Zhang, Yang Yong, Ruihao Gong 외 arxiv

Large Vision-Language Models (VLMs) exhibit impressive multi-modal capabilities but suffer from prohibitive computational and memory demands, due to their long visual token sequences and massive parameter sizes. To addre…

Model Compression

Extreme Compression for Pre-trained Transformers Made Simple and Efficient

2022-06-04 · Xiaoxia Wu, Zhewei Yao, Minjia Zhang, Conglong Li 외

Extreme compression, particularly ultra-low bit precision (binary/ternary) quantization, has been proposed to fit large NLP models on resource-constraint devices. However, to preserve the accuracy for such aggressive com…

Knowledge DistillationQuantization

Extreme Compression of Large Language Models via Additive Quantization

2024-01-11 · Vage Egiazarian, Andrei Panferov, Denis Kuznedelev, Elias Frantar 외

The emergence of accurate open large language models (LLMs) has led to a race towards performant quantization techniques which can enable their execution on end-user devices. In this paper, we revisit the problem of "ext…

CPUGPUInformation RetrievalQuantization