使用pytorch+transformers复现了SimCSE论文中的有监督训练和无监督训练方法

Related tags

Text Data & NLPSimCSE
Overview

SimCSE复现

项目描述

SimCSE是一种简单但是很巧妙的NLP对比学习方法,创新性地引入Dropout的方式,对样本添加噪声,从而达到对正样本增强的目的。 该框架的训练目的为:对于batch中的每个样本,拉近其与正样本之间的距离,拉远其与负样本之间的距离,使得模型能够在大规模无监督语料(也可以使用有监督的语料)中学习到文本相似关系。 详见论文:Simple Contrastive Learning of Sentence EmbeddingsSimCSE官方代码仓库

本项目使用pytorch+transformers复现了SimCSE论文中的有监督训练和无监督训练方法,并且在STS-B数据集上进行消融实验,评价指标为Spearman相关系数,预训练模型为Bert-base-uncased, 验证了SimCSE的有效性。在STS-B数据集上,有监督训练和无监督训练的复现效果如下表。

在无监督训练中,dropout=0.1时,复现效果比原文略差,但也比较接近。当dropout=0.2时,复现效果比原文略高。 ** 但在有监督训练中,不知是否由于batch size过小(原论文使用512),复现效果与论文的效果相差较远,后续会进行排查。 **

训练方法 learning rate batch size dropout Spearman’s correlation
原论文 无监督 3e-5 64 0.1 0.763
复现 无监督 3e-5 64 0.2 0.771
复现 无监督 3e-5 64 0.1 0.748
原论文 有监督 5e-5 512 0.1 0.816
复现 有监督 5e-5 64 0.1 0.764

运行环境

python==3.6、transformers==3.1.0、torch==1.6.0

项目结构

  • data:存放训练数据
    • stsbenchmark:STS-B数据集
      • sts-dev.csv:STS-B验证集
      • sts-test.csv:STS-B验测试集
    • nli_for_simcse.csv:数量275601为的NLI数据集
    • wiki1m_for_simcse.txt:维基百科上获取的100w的文本
  • output:输出目录
  • pretrain_model:预训练模型存放位置
  • script:脚本存放位置。
  • dataset.py
  • model.py:模型代码,包含有监督和无监督损失函数的计算方式
  • train.py:训练代码

使用方法

Quick Start

下载训练数据:

bash script/download_nli.sh
bash script/download_wiki.sh

无监督训练,运行脚本

bash script/run_unsup_train.sh

有监督训练,运行脚本

bash script/run_sup_train.sh

实验

无监督训练

从前四条实验数据中可以看到,较大的batch size在一定程度上可以增加模型的泛化性。

dropout为0.2的时候,训练效果比0.1与0.3更好,有可能dropout=0.1加入的噪声过小,而dropout=0.3加入的噪声过大,增强得到的样本与原始样本差异较大。

learning rate batch size dropout 在哪一步得到best checkpoint 验证集上的得分 测试集上的得分
3e-5 256 0.1 6000 0.800 0.761
3e-5 128 0.1 4200 0.799 0.747
3e-5 64 0.1 10900 0.803 0.748
3e-5 32 0.1 21300 0.787 0.714
3e-5 64 0.2 11200 0.811 0.771
3e-5 64 0.3 6300 0.781 0.745
1e-5 64 0.1 16400 0.798 0.751

有监督训练

有监督实验的复现结果未达到预期,超参数相同时,在验证集上的得分略高于无监督,但是在测试集上,得分基本没有差异。增大有监督训练的学习率,有监督的训练的得分略高于无监督训练, 但还是与论文声称的0.816相差较远,原论文使用512的batch size, 不知是否由于batch size的设置有关,后续会对有监督的训练代码进一步排查。

不过从训练曲线可以看到,有监督训练的收敛速度明显快于无监督训练,这也符合我们的认知。

训练方法 learning rate batch size dropout 在哪一步得到best checkpoint 验证集上的得分 测试集上的得分
无监督 3e-5 64 0.1 10900 0.803 0.748
有监督 3e-5 64 0.1 200 0.810 0.748
有监督 5e-5 64 0.1 2300 0.809 0.764
有监督 3e-5 32 0.1 200 0.808 0.743
有监督 5e-5 32 0.1 200 0.806 0.746

无监督训练过程中,验证集得分的变化曲线: avatar

有监督训练过程中,验证集得分的变化曲线: avatar

REFERENCE

TODO

  • 排查有监督学习的效果不符合预期的原因
ACL22 paper: Imputing Out-of-Vocabulary Embeddings with LOVE Makes Language Models Robust with Little Cost

Imputing Out-of-Vocabulary Embeddings with LOVE Makes Language Models Robust with Little Cost LOVE is accpeted by ACL22 main conference as a long pape

Lihu Chen 32 Jan 03, 2023
Disfl-QA: A Benchmark Dataset for Understanding Disfluencies in Question Answering

Disfl-QA is a targeted dataset for contextual disfluencies in an information seeking setting, namely question answering over Wikipedia passages. Disfl-QA builds upon the SQuAD-v2 (Rajpurkar et al., 2

Google Research Datasets 52 Jun 21, 2022
🐍💯pySBD (Python Sentence Boundary Disambiguation) is a rule-based sentence boundary detection that works out-of-the-box.

pySBD: Python Sentence Boundary Disambiguation (SBD) pySBD - python Sentence Boundary Disambiguation (SBD) - is a rule-based sentence boundary detecti

Nipun Sadvilkar 549 Jan 06, 2023
A fast, efficient universal vector embedding utility package.

Magnitude: a fast, simple vector embedding utility library A feature-packed Python package and vector storage file format for utilizing vector embeddi

Plasticity 1.5k Jan 02, 2023
A Python wrapper for simple offline real-time dictation (speech-to-text) and speaker-recognition using Vosk.

Simple-Vosk A Python wrapper for simple offline real-time dictation (speech-to-text) and speaker-recognition using Vosk. Check out the official Vosk G

2 Jun 19, 2022
Universal Adversarial Triggers for Attacking and Analyzing NLP (EMNLP 2019)

Universal Adversarial Triggers for Attacking and Analyzing NLP This is the official code for the EMNLP 2019 paper, Universal Adversarial Triggers for

Eric Wallace 248 Dec 17, 2022
🎐 a python library for doing approximate and phonetic matching of strings.

jellyfish Jellyfish is a python library for doing approximate and phonetic matching of strings. Written by James Turk James Turk 1.8k Dec 21, 2022

Use fastai-v2 with HuggingFace's pretrained transformers

FastHugs Use fastai v2 with HuggingFace's pretrained transformers, see the notebooks below depending on your task: Text classification: fasthugs_seq_c

Morgan McGuire 111 Nov 16, 2022
HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis

HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis Jungil Kong, Jaehyeon Kim, Jaekyoung Bae In our paper, we p

Jungil Kong 1.1k Jan 02, 2023
Ongoing research training transformer language models at scale, including: BERT & GPT-2

What is this fork of Megatron-LM and Megatron-DeepSpeed This is a detached fork of https://github.com/microsoft/Megatron-DeepSpeed, which in itself is

BigScience Workshop 316 Jan 03, 2023
Compute distance between sequences. 30+ algorithms, pure python implementation, common interface, optional external libs usage.

TextDistance TextDistance -- python library for comparing distance between two or more sequences by many algorithms. Features: 30+ algorithms Pure pyt

Life4 3k Jan 06, 2023
BMInf (Big Model Inference) is a low-resource inference package for large-scale pretrained language models (PLMs).

BMInf (Big Model Inference) is a low-resource inference package for large-scale pretrained language models (PLMs).

OpenBMB 377 Jan 02, 2023
This repository contains the code, models and datasets discussed in our paper "Few-Shot Question Answering by Pretraining Span Selection"

Splinter This repository contains the code, models and datasets discussed in our paper "Few-Shot Question Answering by Pretraining Span Selection", to

Ori Ram 88 Dec 31, 2022
Knowledge Graph,Question Answering System,基于知识图谱和向量检索的医疗诊断问答系统

Knowledge Graph,Question Answering System,基于知识图谱和向量检索的医疗诊断问答系统

wangle 823 Dec 28, 2022
TFIDF-based QA system for AIO2 competition

AIO2 TF-IDF Baseline This is a very simple question answering system, which is developed as a lightweight baseline for AIO2 competition. In the traini

Masatoshi Suzuki 4 Feb 19, 2022
BROS: A Pre-trained Language Model Focusing on Text and Layout for Better Key Information Extraction from Documents

BROS (BERT Relying On Spatiality) is a pre-trained language model focusing on text and layout for better key information extraction from documents. Given the OCR results of the document image, which

Clova AI Research 94 Dec 30, 2022
GooAQ 🥑 : Google Answers to Google Questions!

This repository contains the code/data accompanying our recent work on long-form question answering.

AI2 112 Nov 06, 2022
A Facebook Messenger Chatbot using NLP

A Facebook Messenger Chatbot using NLP This project is about creating a messenger chatbot using basic NLP techniques and models like Logistic Regressi

6 Nov 20, 2022
Natural Language Processing Best Practices & Examples

NLP Best Practices In recent years, natural language processing (NLP) has seen quick growth in quality and usability, and this has helped to drive bus

Microsoft 6.1k Dec 31, 2022
CrossNER: Evaluating Cross-Domain Named Entity Recognition (AAAI-2021)

CrossNER is a fully-labeled collected of named entity recognition (NER) data spanning over five diverse domains (Politics, Natural Science, Music, Literature, and Artificial Intelligence) with specia

Zihan Liu 89 Nov 10, 2022