otto-de / tron

⚡️ Implementation of TRON: Transformer Recommender using Optimized Negative-sampling, accepted at ACM RecSys 2023.

TRON: Scalable Session-Based Transformer Recommender using Optimized Negative Sampling

TRON is a scalable session-based Transformer Recommender using Optimized Negative-sampling. This repository contains the official PyTorch Lightning implementation for our paper: Scaling Session-Based Transformer Recommendations using Optimized Negative Sampling and Loss Functions, authored by Timo Wilm, Philipp Normann, Sophie Baumeister, and Paul-Vincent Kobow.

🎯 Abstract

This work introduces TRON, a scalable session-based Transformer Recommender using Optimized Negative-sampling. Motivated by the scalability and performance limitations of prevailing models such as SASRec and GRU4Rec+, TRON integrates top-k negative sampling and listwise loss functions to enhance its recommendation accuracy. Evaluations on relevant large-scale e-commerce datasets show that TRON improves upon the recommendation quality of current methods while maintaining training speeds similar to SASRec. A live A/B test yielded an 18.14% increase in click-through rate over SASRec, highlighting the potential of TRON in practical settings. For further research, we provide access to our source code and an anonymized dataset.

Offline evaluation results on our private OTTO dataset used for the online A/B test of our three groups.	Online results of our A/B test relative to the SASRec baseline. The error bars indicate the 95% confidence interval.

🌟 RecSys 2023 Research Showcase

Below is our poster presented at the 17th ACM Conference on Recommender Systems (RecSys 2023), which provides a comprehensive overview of the key concepts, methodologies, and results of our TRON research.

🚀 Quick Start

1. Clone the repository:

git clone https://github.com/otto-de/TRON.git

2. Install the dependencies:

pip install pipenv
pipenv install --dev

3. Execute the test scripts:

pipenv run pytest

4. Install 7zip, gzip, and zip utilities on your system:

sudo apt-get install 7zip gzip unzip

5. Prepare a dataset (e.g., yoochoose):

./prepare.sh yoochoose

6. Run the main script with a configuration file:

pipenv run python -m src --config-filename tron/yoochoose.json

🗂️ Repository Structure

.
├── Pipfile
├── Pipfile.lock
├── README.md
├── configs                # Contains experiment configuration files
├── doc                    # Contains the paper and related files
├── prepare.sh             # Script to prepare datasets
├── src                    # Source code for the models
└── test                   # Test scripts

⚙️ Config File Documentation

The config folder contains JSON configuration files for all experiments performed in our research. These configurations detail the model's parameters and options.

Here's an explanation of each parameter in the config file:

• model: The base model to be used (e.g., "sasrec", "gru4rec").
• dataset: The dataset to be used for training (e.g., "yoochoose", "otto", "diginetica").
• hidden_size: The size of the hidden layers and item embeddings.
• num_layers: The number of layers in the model.
• dropout: The dropout rate applied to the model's layers.
• num_batch_negatives: The number of negative samples from the batch. Limited by batch_size - 1.
• num_uniform_negatives: The number of uniformly sampled negatives.
• reject_uniform_session_items: If true, items from the same session won't be used as uniform negatives. Becomes slow if num_uniform_negatives is large.
• reject_in_batch_items: If true, items from the same session won't be used as batch negatives.
• sampling_style: The style of negative sampling to use (e.g., "eventwise", "sessionwise", "batchwise"). Has significant impact on training speed.
• loss: The loss function to use (e.g., "bce", "bpr-max", "ssm").
• bpr_penalty: The penalty factor for BPR-Max loss. Ignored if not using BPR-Max loss.
• max_epochs: The maximum number of training epochs.
• batch_size: The batch size used for training and validation.
• max_session_length: The maximum length of a session. Longer sessions will be truncated.
• lr: The learning rate for the optimizer.
• limit_val_batches: The fraction of validation data to use for the validation step.
• accelerator: The device type to be used for training (e.g., "gpu", "cpu").
• overfit_batches: The fraction or number of batches of training data to use for overfitting. Set to 0 for no overfitting. See PyTorch Lightning docs for more details.
• share_embeddings: If true, the embedding weights are shared between the input and output layers.
• output_bias: If true, includes bias in the output layer.
• shuffling_style: The style of shuffling to use for the training dataset (e.g., "no_shuffling", "shuffling_with_replacement", "shuffling_without_replacement").
• optimizer: The optimizer to use for training (e.g., "adam", "adagrad")
• topk_sampling: If true, top-k negative sampling is enabled.
• topk_sampling_k: If topk_sampling is true, this parameter specifies the number of top k negative samples to be used for training.

Example Config File for TRON on the OTTO Dataset

{
    "model": "sasrec",
    "dataset": "otto",
    "hidden_size": 200,
    "num_layers": 2,
    "dropout": 0.05,
    "num_batch_negatives": 127,
    "num_uniform_negatives": 16384,
    "reject_uniform_session_items": false,
    "reject_in_batch_items": true,
    "sampling_style": "batchwise",
    "topk_sampling": true,
    "topk_sampling_k": 100,
    "loss": "ssm",
    "max_epochs": 10,
    "batch_size": 128,
    "max_session_length": 50,
    "lr": 0.0005,
    "limit_val_batches": 1.0,
    "accelerator": "gpu",
    "overfit_batches": 0,
    "share_embeddings": true,
    "output_bias": false,
    "shuffling_style": "no_shuffling",
    "optimizer": "adam"
}

For all config files used in our experiments, refer to the configs directory.

🙌 Contribution

Contributions to TRON are welcome and appreciated. For issues or suggestions for improvements, please open an issue or create a pull request. We believe that open source knowledge sharing is the best way to advance the field of recommender systems.

📖 Citing

If TRON aids your research, please consider citing our work:

@inproceedings{wilm2023tron,
  title={Scaling Session-Based Transformer Recommendations using Optimized Negative Sampling and Loss Functions},
  author={Wilm, Timo and Normann, Philipp and Baumeister, Sophie and Kobow, Paul-Vincent},
  booktitle={Proceedings of the 17th ACM Conference on Recommender Systems},
  doi = {10.1145/3604915.3610236},
  pages={1023--1026},
  year={2023}
}

📜 License

This project is MIT licensed.

📞 Contact

For any queries or questions, please reach out to us via our LinkedIn profiles:

• Timo Wilm
• Philipp Normann
• Sophie Baumeister

For specific issues related to the codebase or for feature requests, please create a new issue on our GitHub page.

If this project aids your research or you find it interesting, we would appreciate it if you could star ⭐ the repository. Thanks for your support!

tron's Issues

After training the OTTO dataset，how I should deploy the model for requests？

Thank you！