Code for CPM-2 Pre-Train

Last update: Dec 28, 2022

Related tags

Deep Learning CPM-2-Pretrain

Overview

CPM-2 Pre-Train

Pre-train CPM-2 此分支为110亿非 MoE 模型的预训练代码，MoE 模型的预训练代码请切换到 moe 分支

CPM-2技术报告请参考link。

0 模型下载

请在智源资源下载页面进行申请，文件介绍如下：

文件名	描述	参数大小
100000.tar	纯中文模型	110亿
36000.tar	中英文双语模型	110亿
300000.tar	中英文MoE模型	1980亿

1 安装

可以直接拉取我们提供的 Docker 环境：

docker pull gyxthu17/cpm-2:1.0

2 数据

scripts/gen_data.sh 中给出了生成数据文件的脚本示例。该脚本将一个多行的纯文本文件（一个 document 一行）转化为二进制文件（会输出三个 .bin 和三个 .idx 文件），方便模型读取。

3 训练

首先需要将 WORKING_DIR 变量换成 CPM-2 目录的所在路径。调整 NUM_WORKERS 和 NUM_GPUS_PER_WORKER 指定机器数量与每台机器的 GPU 设备数量。修改 ${WORKING_DIR}/src/configs/host_files/hostfile-cpm2 文件将其中的主机名称替换成每台机器的 IP 地址或者和 IP 地址相关联的主机名称。

运行命令：

cd src
bash scripts/pretrain_enc_dec.sh

4 引用

如果您使用了我们的代码，请您引用下面的文章。

@article{cpm-v2,
  title={CPM-2: Large-scale Cost-efficient Pre-trained Language Models},
  author={Zhang, Zhengyan and Gu, Yuxian and Han, Xu and Chen, Shengqi and Xiao, Chaojun and Sun, Zhenbo and Yao, Yuan and Qi, Fanchao and Guan, Jian and Ke, Pei and Cai, Yanzheng and Zeng, Guoyang and Tan, Zhixing and Liu, Zhiyuan and Huang, Minlie and Han, Wentao and Liu, Yang and Zhu, Xiaoyan and Sun, Maosong},
  year={2021}
}

Code for CPM-2 Pre-Train

Related tags

Overview

CPM-2 Pre-Train

0 模型下载

1 安装

2 数据

3 训练

4 引用

Owner

Tsinghua AI

Efficient Sharpness-aware Minimization for Improved Training of Neural Networks

Reverse engineering recurrent neural networks with Jacobian switching linear dynamical systems

This is the repository for the paper "Have I done enough planning or should I plan more?"

Replication Package for AequeVox:Automated Fariness Testing for Speech Recognition Systems

(Py)TOD: Tensor-based Outlier Detection, A General GPU-Accelerated Framework

LightNet++: Boosted Light-weighted Networks for Real-time Semantic Segmentation

How to Become More Salient? Surfacing Representation Biases of the Saliency Prediction Model

Object detection evaluation metrics using Python.

Learn about Spice.ai with in-depth samples

Direct design of biquad filter cascades with deep learning by sampling random polynomials.

Have you ever wondered how cool it would be to have your own A.I

NasirKhusraw - The TSP solved using genetic algorithm and show TSP path overlaid on a map of the Iran provinces & their capitals.

Systematic generalisation with group invariant predictions

Space Invaders For Python

KE-Dialogue: Injecting knowledge graph into a fully end-to-end dialogue system.

Models Supported: AlbUNet [18, 34, 50, 101, 152] (1D and 2D versions for Single and Multiclass Segmentation, Feature Extraction with supports for Deep Supervision and Guided Attention)

Learning to Initialize Neural Networks for Stable and Efficient Training

DiscoNet: Learning Distilled Collaboration Graph for Multi-Agent Perception [NeurIPS 2021]

Repo for the ACMMM20 submission: "Personalized breath based biometric authentication with wearable multimodality".

Implementation of Sequence Generative Adversarial Nets with Policy Gradient