E3D-LSTM - GitLite

KO | EN

gitlite — search

#python #hacktoberfest #ai #llm #golang #rust #deep-learning #linux #awesome #pytorch #security #mcp #ios #open-source

E3D-LSTM

(★ 127)

Unofficial PyTorch implementation of E3D-LSTM

File Explorer

Download Latest Version (.zip)

- .gitkeep
- __init__.py
- dataset.py
- e3d_lstm.py
- trainer.py
- utils.py
Pipfile
README.md
taxibj.png

# Installation Guide

1. Get the code

git clone https://github.com/metrofun/E3D-LSTM

Downloads the entire project code from GitHub to your computer.

cd E3D-LSTM

Moves into the project folder you just downloaded.

2. Python

Easy Recommended

Prerequisites

Git Needed to download the project code from GitHub.
Python 3 On Windows, be sure to check 'Add Python to PATH' during installation.

pipenv install

Type this command into your terminal and run it.

python <실행할 파일명>.py # README에서 정확한 실행 파일명을 확인하세요

Runs the Python script (or module).

✅ If it runs without errors and prints output in the terminal, it worked.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Eidetic 3D LSTM in PyTorch

This is an unofficial and partial PyTorch implementation of "Eidetic 3D LSTM: A Model for Video Prediction and Beyond" [1]

Implementeds E3D-LSTM and a trainer for traffic flow prediction on TaxiBJ dataset[2]

<img src="./taxibj.png" width="300" />

## Modifications
* By default uses a cheaper "Scaled Dot-Product"[3] attention.
* Adds more "LayerNorm"[4] for faster training.

## Installation

1. [Download](https://github.com/lucktroy/DeepST/tree/master/data/TaxiBJ) TaxiBJ[2] dataset into `./data/` folder.
2. Install dependencies from `Pipfile`. By default installs CPU-only Pytorch.

## Usage

```bash
python src/trainer.py
```

## Todo
* [ ] Fix TODOs
* [ ] Do qualitative verification.
* [ ] Introduce configs
* [ ] Add visuals.

## References
[1] Y Wang, L Jiang, MH Yang, LJ Li, M Long, L Fei-Fei. Eidetic 3D LSTM: A Model for Video Prediction and Beyond.  
[2] Junbo Zhang, Yu Zheng, Dekang Qi. Deep Spatio-Temporal Residual Networks for Citywide Crowd Flows Prediction.  
[3] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin. Attention is all you need.  
[4] J. L. Ba, J. R. Kiros, and G. E. Hinton. Layer normalization.