Chinese Pre-Trained Language Models (CPM-LM) Version-I

Overview

CPM-Generate

为了促进中文自然语言处理研究的发展,本项目提供了 CPM-LM (2.6B) 模型的文本生成代码,可用于文本生成的本地测试,并以此为基础进一步研究零次学习/少次学习等场景。[项目首页] [模型下载] [技术报告]

若您想使用CPM-1进行推理,我们建议使用高效推理工具BMInf,支持1060以上显卡单卡推理。

安装

首先安装pytorch等基础依赖,再安装APEX以支持fp16:

pip install -r requirements.txt
git clone https://github.com/NVIDIA/apex
cd apex
pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./

考虑apex的安装容易发生问题,我们构建了对应的Docker容器,可以进行快速环境搭建。安装方式如下:

docker pull dmye/cpm:v0

参考运行指令如下:

:/CPM --name=cpm cpm:v0 ">
sudo docker run --gpus '"device=0,1"' -it -v 
   
    :/CPM  --name=cpm  cpm:v0

   

其中 为代码所在目录,-v进行文件目录挂载

注:感谢qhduan同学提供了基于TensorFlow的使用代码,用作Pytorch之外的备选。

模型

模型下载后文件夹的目录结构需设置如下:

.
├── 80000
│   ├── mp_rank_00_model_states.pt
│   └── mp_rank_01_model_states.pt
└── latest_checkpointed_iteration.txt

为保证下载文件的正确性,文件的checksum如下:

SHA1
71d6b6ad4f47b46724eb82c05da8fb9175e62a7d  80000/mp_rank_00_model_states.pt
42aa247a262e2011fa5e276f1a8389fad6d80edc  80000/mp_rank_01_model_states.pt
MD5
f3f6d2f7d84c6a45290a31dabf79ddac  80000/mp_rank_00_model_states.pt
b0e960be4b5226e759ae6fc5246f9160  80000/mp_rank_01_model_states.pt

使用

提供了命令行交互式生成:

bash scripts/generate_text.sh /path/to/CPM

如不使用交互式输入,可增加第二个参数,告知输入文本的位置

bash scripts/generate_text.sh /path/to/CPM example.txt

运行该脚本需要两块GPU,每张卡的GPU内存占用约为7GB。该项目主要基于 Megatron-LM 进行修改。模型的主体架构与GPT-2一致。

默认的模型并行参数为2,如果需要修改,可以使用change_mp.py,并调整generate_text.sh中的MPSIZEchange_mp.py的使用示例如下:

python change_mp.py /path/to/CPM MPSIZE

这里的/path/to/CPM为模型路径,MPSIZE为一个整数,可以为1或者2的倍数,结果会生成一个新的模型,存储路径为/path/to/CPM_MPSIZE

Tokenization

Tokenization实现主要在data_util/tokenization_gpt2.py,先对于文本进行分词,再使用 SentencePiece 得到 BPE 的结果。由于 SentencePiece 不能有效编码空格和换行符,在 BPE 之前,我们将文本中的空格和换行符替换为\u2582\u2583。生成文本的时候也会对应的把生成的\u2582\u2583替换回空格和换行符。

对应问题已解决。

分类任务零次学习(Zero-shot Learning)

提供了三个任务的零次学习任务脚本以供参考,包括OCNLI、TNEWS和IFLYTEK,数据下载链接。脚本使用方法如下:

# OCNLI
bash scripts/zero-shot-ocnli.sh /path/to/CPM /path/to/dataset
# TNEWS
bash scripts/zero-shot-tnews.sh /path/to/CPM /path/to/dataset
# IFLYTEK
bash scripts/zero-shot-iflytek.sh /path/to/CPM /path/to/dataset

TODO

  • 实验环境的docker镜像
  • 提供各个任务具体的使用模板
  • 公开技术报告
  • 模型并行数可动态调整
  • Fine-tune代码
  • 开源实验中使用的小规模模型参数

引用

@article{cpm-v1,
  title={CPM: A Large-scale Generative Chinese Pre-trained Language Model},
  author={Zhang, Zhengyan and Han, Xu, and Zhou, Hao, and Ke, Pei, and Gu, Yuxian and Ye, Deming and Qin, Yujia and Su, Yusheng and Ji, Haozhe and Guan, Jian and Qi, Fanchao and Wang, Xiaozhi and Zheng, Yanan and Zeng, Guoyang and Cao, Huanqi and Chen, Shengqi and Li, Daixuan and Sun, Zhenbo and Liu, Zhiyuan and Huang, Minlie and Han, Wentao and Tang, Jie and Li, Juanzi and Sun, Maosong},
  year={2020}
}
Owner
Tsinghua AI
Tsinghua AI
Spooky Skelly For Python

_____ _ _____ _ _ _ | __| ___ ___ ___ | |_ _ _ | __|| |_ ___ | || | _ _ |__ || . || . || . || '

Kur0R1uka 1 Dec 23, 2021
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
A Chinese to English Neural Model Translation Project

ZH-EN NMT Chinese to English Neural Machine Translation This project is inspired by Stanford's CS224N NMT Project Dataset used in this project: News C

Zhenbang Feng 29 Nov 26, 2022
A PyTorch implementation of the WaveGlow: A Flow-based Generative Network for Speech Synthesis

WaveGlow A PyTorch implementation of the WaveGlow: A Flow-based Generative Network for Speech Synthesis Quick Start: Install requirements: pip install

Yuchao Zhang 204 Jul 14, 2022
PyTorch Language Model for 1-Billion Word (LM1B / GBW) Dataset

PyTorch Large-Scale Language Model A Large-Scale PyTorch Language Model trained on the 1-Billion Word (LM1B) / (GBW) dataset Latest Results 39.98 Perp

Ryan Spring 114 Nov 04, 2022
A Python script that compares files in directories

compare-files A Python script that compares files in different directories, this is similar to the command filecmp.cmp(f1, f2). I made this script in

Colvin 1 Oct 15, 2021
Natural language processing summarizer using 3 state of the art Transformer models: BERT, GPT2, and T5

NLP-Summarizer Natural language processing summarizer using 3 state of the art Transformer models: BERT, GPT2, and T5 This project aimed to provide in

Samuel Sharkey 1 Feb 07, 2022
Pytorch version of BERT-whitening

BERT-whitening This is the Pytorch implementation of "Whitening Sentence Representations for Better Semantics and Faster Retrieval". BERT-whitening is

Weijie Liu 255 Dec 27, 2022
ChainKnowledgeGraph, 产业链知识图谱包括A股上市公司、行业和产品共3类实体

ChainKnowledgeGraph, 产业链知识图谱包括A股上市公司、行业和产品共3类实体,包括上市公司所属行业关系、行业上级关系、产品上游原材料关系、产品下游产品关系、公司主营产品、产品小类共6大类。 上市公司4,654家,行业511个,产品95,559条、上游材料56,824条,上级行业480条,下游产品390条,产品小类52,937条,所属行业3,946条。

liuhuanyong 415 Jan 06, 2023
Pipeline for fast building text classification TF-IDF + LogReg baselines.

Text Classification Baseline Pipeline for fast building text classification TF-IDF + LogReg baselines. Usage Instead of writing custom code for specif

Dani El-Ayyass 57 Dec 07, 2022
KoBERT - Korean BERT pre-trained cased (KoBERT)

KoBERT KoBERT Korean BERT pre-trained cased (KoBERT) Why'?' Training Environment Requirements How to install How to use Using with PyTorch Using with

SK T-Brain 1k Jan 02, 2023
Code for the paper "Flexible Generation of Natural Language Deductions"

Code for the paper "Flexible Generation of Natural Language Deductions"

Kaj Bostrom 12 Nov 11, 2022
**NSFW** A chatbot based on GPT2-chitchat

DangBot -- 好怪哦,再来一句 卡群怪话bot,powered by GPT2 for Chinese chitchat Training Example: python train.py --lr 5e-2 --epochs 30 --max_len 300 --batch_size 8

Tommy Yang 11 Jul 21, 2022
An easy to use, user-friendly and efficient code for extracting OpenAI CLIP (Global/Grid) features from image and text respectively.

Extracting OpenAI CLIP (Global/Grid) Features from Image and Text This repo aims at providing an easy to use and efficient code for extracting image &

Jianjie(JJ) Luo 13 Jan 06, 2023
A minimal Conformer ASR implementation adapted from ESPnet.

Conformer ASR A minimal Conformer ASR implementation adapted from ESPnet. Introduction I want to use the pre-trained English ASR model provided by ESP

Niu Zhe 3 Jan 24, 2022
Various Algorithms for Short Text Mining

Short Text Mining in Python Introduction This package shorttext is a Python package that facilitates supervised and unsupervised learning for short te

Kwan-Yuet 466 Dec 06, 2022
The Sudachi synonym dictionary in Solar format.

solr-sudachi-synonyms The Sudachi synonym dictionary in Solar format. Summary Run a script that checks for updates to the Sudachi dictionary every hou

Karibash 3 Aug 19, 2022
The Internet Archive Research Assistant - Daily search Internet Archive for new items matching your keywords

The Internet Archive Research Assistant - Daily search Internet Archive for new items matching your keywords

Kay Savetz 60 Dec 25, 2022
XLNet: Generalized Autoregressive Pretraining for Language Understanding

Introduction XLNet is a new unsupervised language representation learning method based on a novel generalized permutation language modeling objective.

Zihang Dai 6k Jan 07, 2023
Japanese Long-Unit-Word Tokenizer with RemBertTokenizerFast of Transformers

Japanese-LUW-Tokenizer Japanese Long-Unit-Word (国語研長単位) Tokenizer for Transformers based on 青空文庫 Basic Usage from transformers import RemBertToken

Koichi Yasuoka 3 Dec 22, 2021