When BERT Plays the Lottery, All Tickets Are Winning

Overview

When BERT Plays the Lottery, All Tickets Are Winning

Large Transformer-based models were shown to be reducible to a smaller number of self-attention heads and layers. We consider this phenomenon from the perspective of the lottery ticket hypothesis, using both structured and magnitude pruning. For fine-tuned BERT, we show that (a) it is possible to find subnetworks achieving performance that is comparable with that of the full model, and (b) similarly-sized subnetworks sampled from the rest of the model perform worse. Strikingly, with structured pruning even the worst possible subnetworks remain highly trainable, indicating that most pre-trained BERT weights are potentially useful. We also study the "good" subnetworks to see if their success can be attributed to superior linguistic knowledge, but find them unstable, and not explained by meaningful self-attention patterns.

Environment

Install the requirements in your python 3.7.7 virtual environment.

pip install -r requirements.txt

These experiments were done on multi-gpu environment, were some experiments, benchmarks were run parallel. So some changes to the bash scripts to make it work for your environment.

Dataset

  1. Download the GLUE dataset using data/download_glue.py and data/download_mnli_data.py. Follow the instructions in data/download_glue.py docstring for MRPC.
  2. All data for the tasks should be organized in data/glue/task_name/ structure.
  3. Extract the attention pattern classification labelled data.
    cd data
    tar -xvf head_classification_data.tar.gz

Training, Masking, and Evaluation

Switch cwd to src (cd src) as many paths are relative from that directory.

  1. Fine-tune the BERT on GLUE tasks
./train.sh
  1. Obtain the masks
./find_masks.sh
  1. Train models with the masks applied in good, random and bad settings.
./train_with_masks.sh
  1. Evaluate the trained models
./evaluate.sh

Note: These experiments were run through course of time and now stiched together into single scripts. So it might be better to run the training and evaluation commands in them one by one.

  1. Train the CNN classifier on attention patterns normed and raw.
python classify_attention_patterns.py
python classify_normed_patterns.py

These only train the classifier.

Evaluation Analysis and Final Results

These are primarily done in jupyter notebooks in experiment_analysis directory. There are many experimental notebooks there. Here are the important ones used to generate results included in the paper.

  1. Importance pruning Heatmaps. Ignore the final "train_subset" and "hans" settings.
  2. Magnitude pruning Heatmap
  3. Overlap of surviving components
  4. Generate the random baseline
  5. Attention Classification Patterns
  6. Evaluation Result Comparisons and table
  7. Statistics on mask correlation across seeds
Owner
Sai
Machine Learning Researcher - NLP
Sai
VR-Caps: A Virtual Environment for Active Capsule Endoscopy

VR-Caps: A Virtual Environment for Capsule Endoscopy Overview We introduce a virtual active capsule endoscopy environment developed in Unity that prov

DeepMIA Lab 90 Dec 27, 2022
This is an official implementation for "SimMIM: A Simple Framework for Masked Image Modeling".

SimMIM By Zhenda Xie*, Zheng Zhang*, Yue Cao*, Yutong Lin, Jianmin Bao, Zhuliang Yao, Qi Dai and Han Hu*. This repo is the official implementation of

Microsoft 674 Dec 26, 2022
RoadMap and preparation material for Machine Learning and Data Science - From beginner to expert.

ML-and-DataScience-preparation This repository has the goal to create a learning and preparation roadMap for Machine Learning Engineers and Data Scien

33 Dec 29, 2022
Pairwise model for commonlit competition

Pairwise model for commonlit competition To run: - install requirements - create input directory with train_folds.csv and other competition data - cd

abhishek thakur 45 Aug 31, 2022
Extending JAX with custom C++ and CUDA code

Extending JAX with custom C++ and CUDA code This repository is meant as a tutorial demonstrating the infrastructure required to provide custom ops in

Dan Foreman-Mackey 237 Dec 23, 2022
Codes for the compilation and visualization examples to the HIF vegetation dataset

High-impedance vegetation fault dataset This repository contains the codes that compile the "Vegetation Conduction Ignition Test Report" data, which a

1 Dec 12, 2021
Multivariate Boosted TRee

Multivariate Boosted TRee What is MBTR MBTR is a python package for multivariate boosted tree regressors trained in parameter space. The package can h

SUPSI-DACD-ISAAC 61 Dec 19, 2022
RuDOLPH: One Hyper-Modal Transformer can be creative as DALL-E and smart as CLIP

[Paper] [Хабр] [Model Card] [Colab] [Kaggle] RuDOLPH 🦌 🎄 ☃️ One Hyper-Modal Tr

Sber AI 230 Dec 31, 2022
Differentiable Quantum Chemistry (only Differentiable Density Functional Theory and Hartree Fock at the moment)

DQC: Differentiable Quantum Chemistry Differentiable quantum chemistry package. Currently only support differentiable density functional theory (DFT)

75 Dec 02, 2022
Individual Tree Crown classification on WorldView-2 Images using Autoencoder -- Group 9 Weak learners - Final Project (Machine Learning 2020 Course)

Created by Olga Sutyrina, Sarah Elemili, Abduragim Shtanchaev and Artur Bille Individual Tree Crown classification on WorldView-2 Images using Autoenc

2 Dec 08, 2022
Towhee is a flexible machine learning framework currently focused on computing deep learning embeddings over unstructured data.

Towhee is a flexible machine learning framework currently focused on computing deep learning embeddings over unstructured data.

1.7k Jan 08, 2023
Content shared at DS-OX Meetup

Streamlit-Projects Streamlit projects available in this repo: An introduction to Streamlit presented at DS-OX (Feb 26, 2020) meetup Streamlit 101 - Ja

Arvindra 69 Dec 23, 2022
Reproduce ResNet-v2(Identity Mappings in Deep Residual Networks) with MXNet

Reproduce ResNet-v2 using MXNet Requirements Install MXNet on a machine with CUDA GPU, and it's better also installed with cuDNN v5 Please fix the ran

Wei Wu 531 Dec 04, 2022
GANSketchingJittor - Implementation of Sketch Your Own GAN in Jittor

GANSketching in Jittor Implementation of (Sketch Your Own GAN) in Jittor(计图). Or

Bernard Tan 10 Jul 02, 2022
Zero-Shot Text-to-Image Generation VQGAN+CLIP Dockerized

VQGAN-CLIP-Docker About Zero-Shot Text-to-Image Generation VQGAN+CLIP Dockerized This is a stripped and minimal dependency repository for running loca

Kevin Costa 73 Sep 11, 2022
PyTorch CZSL framework containing GQA, the open-world setting, and the CGE and CompCos methods.

Compositional Zero-Shot Learning This is the official PyTorch code of the CVPR 2021 works Learning Graph Embeddings for Compositional Zero-shot Learni

EML Tübingen 70 Dec 27, 2022
Implementation of Google Brain's WaveGrad high-fidelity vocoder

WaveGrad Implementation (PyTorch) of Google Brain's high-fidelity WaveGrad vocoder (paper). First implementation on GitHub with high-quality generatio

Ivan Vovk 363 Dec 27, 2022
Text Generation by Learning from Demonstrations

Text Generation by Learning from Demonstrations The README was last updated on March 7, 2021. The repo is based on fairseq (v0.9.?). Paper arXiv Prere

38 Oct 21, 2022
A flexible framework of neural networks for deep learning

Chainer: A deep learning framework Website | Docs | Install Guide | Tutorials (ja) | Examples (Official, External) | Concepts | ChainerX Forum (en, ja

Chainer 5.8k Jan 06, 2023
Selfplay In MultiPlayer Environments

This project allows you to train AI agents on custom-built multiplayer environments, through self-play reinforcement learning.

200 Jan 08, 2023