The official implementation of You Only Compress Once: Towards Effective and Elastic BERT Compression via Exploit-Explore Stochastic Nature Gradient.

Overview

You Only Compress Once: Towards Effective and Elastic BERT Compression via Exploit-Explore Stochastic Nature Gradient (paper)

@misc{zhang2021compress,
      title={You Only Compress Once: Towards Effective and Elastic BERT Compression via Exploit-Explore Stochastic Nature Gradient}, 
      author={Shaokun Zhang and Xiawu Zheng and Chenyi Yang and Yuchao Li and Yan Wang and Fei Chao and Mengdi Wang and Shen Li and Jun Yang and Rongrong Ji},
      year={2021},
      eprint={2106.02435},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
      } 

Overview

This repository is the official implementation of You Only Compress Once: Towards Effective and Elastic BERT Compression via Exploit-Explore Stochastic Nature Gradient

📋 We propose a novel approach, YOCO-BERT, to achieve compress once and deploy everywhere. Compared with state of-the-art algorithms, YOCO-BERT provides more compact models, yet achieving superior average accuracy improvement on the GLUE.

Requirements

  • Python > 3.6
  • Pytorch = 1.7.0
  • transformers = 3.5.0

Training

To train the super-BERTs in the paper, run this command:

python train_superbert.py --cfg /path_to_superbert_training_config/config.yaml

Searching

To search the optimal sub-BERTs given any constraints in the paper, run this command:

python search_subbert.py --cfg /path_to_subbert_searching_config/config.yaml

Evaluation

The evaluation results will be reported after the searching process.

Config

We release all the traning and searching configs in config

Results

Our model achieves the following performance on :

GLUE

Results given various FlOPs and parameters.

Results under common constraints (compress to no more than 66M)

Datasets SST-2 MRPC CoLA RTE MNLI QQP QNLI
Results 92.8 90.3 59.8 72.9 82.6 90.5 87.2

📋 The detailed metrics used in this code are reported in the paper.

Licence

This repository is released under the MIT license. See LICENSE for more information.

Contact

Any problem regarding this code re-implementation, feel free to contact the first author: [email protected]

You might also like...
Trains an agent with stochastic policy gradient ascent to solve the Lunar Lander challenge from OpenAI
Trains an agent with stochastic policy gradient ascent to solve the Lunar Lander challenge from OpenAI

Introduction This script trains an agent with stochastic policy gradient ascent to solve the Lunar Lander challenge from OpenAI. In order to run this

A Momentumized, Adaptive, Dual Averaged Gradient Method for Stochastic Optimization
A Momentumized, Adaptive, Dual Averaged Gradient Method for Stochastic Optimization

MADGRAD Optimization Method A Momentumized, Adaptive, Dual Averaged Gradient Method for Stochastic Optimization pip install madgrad Try it out! A best

A PyTorch implementation of Learning to learn by gradient descent by gradient descent

Intro PyTorch implementation of Learning to learn by gradient descent by gradient descent. Run python main.py TODO Initial implementation Toy data LST

Official implementation of the ICML2021 paper
Official implementation of the ICML2021 paper "Elastic Graph Neural Networks"

ElasticGNN This repository includes the official implementation of ElasticGNN in the paper "Elastic Graph Neural Networks" [ICML 2021]. Xiaorui Liu, W

Pytorch implementation for Patient Knowledge Distillation for BERT Model Compression

Patient Knowledge Distillation for BERT Model Compression Knowledge distillation for BERT model Installation Run command below to install the environm

Parasite: a tool allowing you to compress and decompress files, to reduce their size
Parasite: a tool allowing you to compress and decompress files, to reduce their size

🦠 Parasite 🦠 Parasite is a tool written in Python3 allowing you to "compress" any file, reducing its size. ⭐ Features ⭐ + Fast + Good optimization,

Code for the ICML 2021 paper
Code for the ICML 2021 paper "Bridging Multi-Task Learning and Meta-Learning: Towards Efficient Training and Effective Adaptation", Haoxiang Wang, Han Zhao, Bo Li.

Bridging Multi-Task Learning and Meta-Learning Code for the ICML 2021 paper "Bridging Multi-Task Learning and Meta-Learning: Towards Efficient Trainin

Official implementation of Generalized Data Weighting via Class-level Gradient Manipulation (NeurIPS 2021).
Official implementation of Generalized Data Weighting via Class-level Gradient Manipulation (NeurIPS 2021).

Generalized Data Weighting via Class-level Gradient Manipulation This repository is the official implementation of Generalized Data Weighting via Clas

Pre-trained BERT Models for Ancient and Medieval Greek, and associated code for LaTeCH 2021 paper titled -
Pre-trained BERT Models for Ancient and Medieval Greek, and associated code for LaTeCH 2021 paper titled - "A Pilot Study for BERT Language Modelling and Morphological Analysis for Ancient and Medieval Greek"

Ancient Greek BERT The first and only available Ancient Greek sub-word BERT model! State-of-the-art post fine-tuning on Part-of-Speech Tagging and Mor

Comments
  • Bugs repair please~

    Bugs repair please~

    Hi, buddy, there are a lot of bugs that I cannot run the "train-supernet.py" successfully !

    model settings issue in train_supernet.py

    In addition, there is one imported model >from model.supermodel import BertForSequenceClassification And you try to create teacher_model and super_model with the same SuperModel (SuperEmbedding, SuperLinear, SuperLayerNorm) setting?

    opened by pengfeiZhao1993 3
Owner
null
You Only Hypothesize Once: Point Cloud Registration with Rotation-equivariant Descriptors

You Only Hypothesize Once: Point Cloud Registration with Rotation-equivariant Descriptors In this paper, we propose a novel local descriptor-based fra

Haiping Wang 80 Dec 15, 2022
Code for You Only Cut Once: Boosting Data Augmentation with a Single Cut

You Only Cut Once (YOCO) YOCO is a simple method/strategy of performing augmenta

null 88 Dec 28, 2022
Explore extreme compression for pre-trained language models

Code for paper "Exploring extreme parameter compression for pre-trained language models ICLR2022"

twinkle 16 Nov 14, 2022
Sound and Cost-effective Fuzzing of Stripped Binaries by Incremental and Stochastic Rewriting

StochFuzz: A New Solution for Binary-only Fuzzing StochFuzz is a (probabilistically) sound and cost-effective fuzzing technique for stripped binaries.

Zhuo Zhang 164 Dec 5, 2022
I-BERT: Integer-only BERT Quantization

I-BERT: Integer-only BERT Quantization HuggingFace Implementation I-BERT is also available in the master branch of HuggingFace! Visit the following li

Sehoon Kim 139 Dec 27, 2022
An Image compression simulator that uses Source Extractor and Monte Carlo methods to examine the post compressive effects different compression algorithms have.

ImageCompressionSimulation An Image compression simulator that uses Source Extractor and Monte Carlo methods to examine the post compressive effects o

James Park 1 Dec 11, 2021
Revisiting Discriminator in GAN Compression: A Generator-discriminator Cooperative Compression Scheme (NeurIPS2021)

Revisiting Discriminator in GAN Compression: A Generator-discriminator Cooperative Compression Scheme (NeurIPS2021) Overview Prerequisites Linux Pytho

Shaojie Li 34 Mar 31, 2022
Storchastic is a PyTorch library for stochastic gradient estimation in Deep Learning

Storchastic is a PyTorch library for stochastic gradient estimation in Deep Learning

Emile van Krieken 140 Dec 30, 2022
On the model-based stochastic value gradient for continuous reinforcement learning

On the model-based stochastic value gradient for continuous reinforcement learning This repository is by Brandon Amos, Samuel Stanton, Denis Yarats, a

Facebook Research 46 Dec 15, 2022