EasyTPP [ICLR 2024]

<div align="center"> <a href="PyVersion"> <img alt="Python Version" src="https://img.shields.io/badge/python-3.9+-blue.svg"> </a> <a href="LICENSE-CODE"> <img alt="Code License" src="https://img.shields.io/badge/license-Apache-000000.svg?&color=f5de53"> </a> <a href="commit"> <img alt="Last Commit" src="https://img.shields.io/github/last-commit/ant-research/EasyTemporalPointProcess"> </a> </div> <div align="center"> <a href="https://pypi.python.org/pypi/easy-tpp/"> <img alt="PyPI version" src="https://img.shields.io/pypi/v/easy-tpp.svg?style=flat-square&color=b7534" /> </a> <a href="https://static.pepy.tech/personalized-badge/easy-tpp"> <img alt="Downloads" src="https://static.pepy.tech/personalized-badge/easy-tpp?period=total&units=international_system&left_color=black&right_color=blue&left_text=Downloads" /> </a> <a href="https://huggingface.co/easytpp" target="_blank"> <img alt="Hugging Face" src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-EasyTPP-ffc107?color=ffc107&logoColor=white" /> </a> <a href="https://github.com/ant-research/EasyTemporalPointProcess/issues"> <img alt="Open Issues" src="https://img.shields.io/github/issues-raw/ant-research/EasyTemporalPointProcess" /> </a> </div>

EasyTPP is an easy-to-use development and application toolkit for Temporal Point Process (TPP), with key features in configurability, compatibility and reproducibility. We hope this project could benefit both researchers and practitioners with the goal of easily customized development and open benchmarking in TPP. <span id='top'/>

| <a href='#features'>Features</a> | <a href='#model-list'>Model List</a> | <a href='#dataset'>Dataset</a> | <a href='#quick-start'>Quick Start</a> | <a href='#benchmark'>Benchmark</a> |<a href='#doc'>Documentation</a> |<a href='#todo'>Todo List</a> | <a href='#citation'>Citation</a> |<a href='#acknowledgment'>Acknowledgement</a> | <a href='#star-history'>Star History</a> |

News

<span id='news'/>

• [11-06-2025] We have released a new version of EasyTPP that exclusively supports PyTorch. TensorFlow support has been removed to streamline the codebase and focus on PyTorch-based implementations.
• [11-05-2025] Added the implementation of the S2P2 model, presented at NeurIPS'2025.
• [02-17-2024] EasyTPP supports HuggingFace dataset API: all datasets have been published in HuggingFace Repo and see tutorial notebook for an example of usage.
• [01-16-2024] Our paper EasyTPP: Towards Open Benchmarking Temporal Point Process is accepted by ICLR'2024!

<details> <summary>Click to see previous news</summary> <p> - [09-30-2023] We published two textual event sequence datasets [GDELT](https://drive.google.com/drive/folders/1Ms-ATMMFf6v4eesfJndyuPLGtX58fCnk) and [Amazon-text-review](https://drive.google.com/drive/folders/1-SLYyrl7ucEG7NpSIF0eSoG9zcbZagZw) that are used in our paper [LAMP](https://arxiv.org/abs/2305.16646), where LLM can be applied for event prediction! See [Documentation](https://ant-research.github.io/EasyTemporalPointProcess/user_guide/dataset.html#preprocessed-datasets) for more details. - [09-30-2023] Two of our papers [Language Model Can Improve Event Prediction by Few-Shot Abductive Reasoning](https://arxiv.org/abs/2305.16646) (LAMP) and [Prompt-augmented Temporal Point Process for Streaming Event Sequence](https://arxiv.org/abs/2310.04993) (PromptTPP) are accepted by NeurIPS'2023! - [09-02-2023] We published two non-anthropogenic datasets [earthquake](https://drive.google.com/drive/folders/1ubeIz_CCNjHyuu6-XXD0T-gdOLm12rf4) and [volcano eruption](https://drive.google.com/drive/folders/1KSWbNi8LUwC-dxz1T5sOnd9zwAot95Tp?usp=drive_link)! See <a href='#dataset'>Dataset</a> for details. - [05-29-2023] We released ``EasyTPP`` v0.0.1! - [12-27-2022] Our paper [Bellman Meets Hawkes: Model-Based Reinforcement Learning via Temporal Point Processes](https://arxiv.org/abs/2201.12569) was accepted by AAAI'2023! - [10-01-2022] Our paper [HYPRO: A Hybridly Normalized Probabilistic Model for Long-Horizon Prediction of Event Sequences](https://arxiv.org/abs/2210.01753) was accepted by NeurIPS'2022! - [05-01-2022] We started to develop `EasyTPP`.</p> </details>

Features <a href='#top'>[Back to Top]</a>

<span id='features'/>

• Configurable and customizable: models are modularized and configurable，with abstract classes to support developing customized TPP models.
• PyTorch-based implementation: EasyTPP implements state-of-the-art TPP models using PyTorch 1.7.0+, providing a clean and modern deep learning framework.
• Reproducible: all the benchmarks can be easily reproduced.
• Hyper-parameter optimization: a pipeline of optuna-based HPO is provided.

Model List <a href='#top'>[Back to Top]</a>

<span id='model-list'/>

We provide reference implementations of various state-of-the-art TPP papers:

| No | Publication | Model | Paper | Implementation | |:---:|:-----------:|:-------------:|:-----------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------| | 1 | KDD'16 | RMTPP | Recurrent Marked Temporal Point Processes: Embedding Event History to Vector | PyTorch | | 2 | NeurIPS'17 | NHP | The Neural Hawkes Process: A Neurally Self-Modulating Multivariate Point Process | PyTorch | | 3 | NeurIPS'19 | FullyNN | Fully Neural Network based Model for General Temporal Point Processes | PyTorch | | 4 | ICML'20 | SAHP | Self-Attentive Hawkes process | PyTorch | | 5 | ICML'20 | THP | Transformer Hawkes process | PyTorch | | 6 | ICLR'20 | IntensityFree | Intensity-Free Learning of Temporal Point Processes | PyTorch | | 7 | ICLR'21 | ODETPP | Neural Spatio-Temporal Point Processes (simplified) | PyTorch | | 8 | ICLR'22 | AttNHP | Transformer Embeddings of Irregularly Spaced Events and Their Participants | PyTorch | | 9 | NeurIPS'25 | S2P2 | Deep Continuous-Time State-Space Models for Marked Event Sequences | PyTorch |

Dataset <a href='#top'>[Back to Top]</a>

<span id='dataset'/>

We preprocessed one synthetic and five real world datasets from widely-cited works that contain diverse characteristics in terms of their application domains and temporal statistics:

• Synthetic: a univariate Hawkes process simulated by Tick library.
• Retweet (Zhou, 2013): timestamped user retweet events.
• Taxi (Whong, 2014): timestamped taxi pick-up events.
• StackOverflow (Leskovec, 2014): timestamped user badge reward events in StackOverflow.
• Taobao (Xue et al, 2022): timestamped user online shopping behavior events in Taobao platform.
• Amazon (Xue et al, 2022): timestamped user online shopping behavior events in Amazon platform.

Per users' request, we processed two non-anthropogenic datasets

• Earthquake: timestamped earthquake events over the Conterminous U.S fro