StackRec: Efficient Training of Very Deep Sequential Recommender Models by Iterative Stacking

Last update: May 29, 2022

Related tags

Deep Learning StackRec

Overview

StackRec: Efficient Training of Very Deep Sequential Recommender Models by Iterative Stacking

Datasets

You can download datasets that have been pre-processed:

ML20: https://pan.baidu.com/s/14pk0N-yraoxGgsnbJRPG5Q code(提取码): 7yha
ColdRec: https://pan.baidu.com/s/1AkTImhvnD8WyXCTOuynZ8g code(提取码): 9cs2 https://pan.baidu.com/s/1byW5uCZbdEjGzoXJAlPalQ code(提取码): 856z
Video-6M: https://drive.google.com/file/d/1wd3xzF9VnZ6r35nMb3-H4E31vWK87vjW/view?usp=sharing

We construct a large-scale session-based recommendation dataset (denoted as Video-6M) by collecting the interactiton behaviors of nearly 6 million users in a week from a commercial recommender system. The dataset can be used to evaluate very deep recommendation models (up to 100 layers), such as NextItNet (as shown in our paper StackRec（SIGIR2021)). If you use this dataset in your paper, you should cite our NextItNet and StackRec for publish permission.

@article{yuan2019simple,
	title={A simple convolutional generative network for next item recommendation},
	author={Yuan, Fajie and Karatzoglou, Alexandros and Arapakis, Ioannis and Jose, Joemon M and He, Xiangnan},
	journal={Proceedings of the Twelfth ACM International Conference on Web Search and Data Mining},
	year={2019}
}

@article{wang2020stackrec,
  title={StackRec: Efficient Training of Very Deep Sequential Recommender Models by Iterative Stacking},
  author={Wang, Jiachun and Yuan, Fajie and Chen, Jian and Wu, Qingyao and Li, Chengmin and Yang, Min and Sun, Yang and Zhang, Guoxiao},
  journal={Proceedings of the 44th International ACM SIGIR conference on Research and Development in Information Retrieval},
  year={2021}
}

File Description

requirements.txt: the experiment environment

train_nextitnet_sc1.sh: the shell script to train StackRec with NextItNet in CL scenario
train_nextitnet_sc2.sh: the shell script to train StackRec with NextItNet in TF scenario
train_nextitnet_sc3.sh: the shell script to train StackRec with NextItNet in TS scenario
deep_nextitnet.py: the training file of NextItNet
deep_nextitnet_coldrec.py: the training file of NextItNet customized for coldrec source dataset
data_loader.py: the dataset loading file of NextItNet and GRec
data_loader_finetune.py: the dataset loading file of NextItNet and GRec customized for coldrec dataset
generator_deep.py: the model file of NextItNet
ops.py: the module file of NextItNet and GRec with stacking methods doubling blocks
ops_copytop.py: the module file of NextItNet with stacking methods allowed to stack top blocks
ops_original.py: the module file of NextItNet with stacking methods without alpha
fineall.py: the training file of NextItNet customized for coldrec target dataset

train_grec_sc1.sh: the shell script to train StackRec with GRec in CL scenario
deep_GRec: the training file of GRec
generator_deep_GRec.py: the model file of GRec
utils_GRec.py: some tools for GRec

train_sasrec_sc1.sh: the shell script to train StackRec with SASRec in CL scenario
baseline_SASRec.py: the training file of SASRec
Data_loader_SASRec.py: the dataset loading file of SASRec
SASRec_Alpha.py: the model file of SASRec

train_ssept_sc1.sh: the shell script to train StackRec with SSEPT in CL scenario
baseline_SSEPT.py: the training file of SSEPT
Data_loader_SSEPT.py: the dataset loading file of SSEPT
SSEPT_Alpha.py: the model file of SSEPT
utils.py: some tools for SASRec and SSEPT
Modules.py: the module file of SASRec and SSEPT with stacking methods

Stacking with NextItNet

Train in the CL scenario

Execute example:

sh train_nextitnet_sc1.sh

Train in the TS scenario

Execute example:

sh train_nextitnet_sc2.sh

Train in the TF scenario

Execute example:

sh train_nextitnet_sc3.sh

Stacking with GRec

Execute example:

sh train_grec_sc1.sh

Stacking with SASRec

Execute example:

sh train_sasrec_sc1.sh

Stacking with SSEPT

Execute example:

sh train_ssept_sc1.sh

Key Configuration

method: five stacking methods including from_scratch, stackC, stackA, stackR and stackE
data_ratio: the percentage of training data
dilation_count: the number of dilation factors {1,2,4,8}
num_blocks: the number of residual blocks
load_model: whether load pre-trained model or not

StackRec: Efficient Training of Very Deep Sequential Recommender Models by Iterative Stacking

Related tags

Overview

StackRec: Efficient Training of Very Deep Sequential Recommender Models by Iterative Stacking

Datasets

File Description

Stacking with NextItNet

Train in the CL scenario

Train in the TS scenario

Train in the TF scenario

Stacking with GRec

Stacking with SASRec

Stacking with SSEPT

Key Configuration

Owner

GPU-accelerated PyTorch implementation of Zero-shot User Intent Detection via Capsule Neural Networks

Official PyTorch implementation of "Evolving Search Space for Neural Architecture Search"

BEAMetrics: Benchmark to Evaluate Automatic Metrics in Natural Language Generation

Node-level Graph Regression with Deep Gaussian Process Models

Code and data to accompany the camera-ready version of "Cross-Attention is All You Need: Adapting Pretrained Transformers for Machine Translation" in EMNLP 2021

This framework implements the data poisoning method found in the paper Adversarial Examples Make Strong Poisons

Beginner-friendly repository for Hacktober Fest 2021. Start your contribution to open source through baby steps. 💜

An implementation of paper `Real-time Convolutional Neural Networks for Emotion and Gender Classification` with PaddlePaddle.

Fast, flexible and easy to use probabilistic modelling in Python.

Spam your friends and famly and when you do your famly will disown you and you will have no friends.

Create UIs for prototyping your machine learning model in 3 minutes

GLANet - The code for Global and Local Alignment Networks for Unpaired Image-to-Image Translation arxiv

This is an official implementation for "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows" on Semantic Segmentation.

Optimized primitives for collective multi-GPU communication

QueryDet: Cascaded Sparse Query for Accelerating High-Resolution SmallObject Detection

CVPR 2021

A Broader Picture of Random-walk Based Graph Embedding

JittorVis - Visual understanding of deep learning models

Channel Pruning for Accelerating Very Deep Neural Networks (ICCV'17)

pytorch implementation of "Contrastive Multiview Coding", "Momentum Contrast for Unsupervised Visual Representation Learning", and "Unsupervised Feature Learning via Non-Parametric Instance-level Discrimination"