OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Liquid Time Constant Networks

Liquid Time-Constant Networks is an open-source Python research repository for training continuous-time neural models with backpropagation through time. It includes implementations of liquid time-constant networks, Neural ODEs, continuous-time RNNs and continuous-time GRUs, plus task scripts for sequence data.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Liquid Time Constant Networks — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you want research code and task-specific scripts for training and evaluating continuous-time models on sequence datasets, and can work within the stated Python, TensorFlow and Ubuntu environment.

Who it is for

  • Researchers comparing continuous-time recurrent model families.
  • Practitioners studying sequence tasks such as gesture segmentation, activity recognition, occupancy detection or time-series forecasting.

Documented uses

  • Train a selected recurrent model on one of the repository’s prepared task datasets.
  • Compare validation and test metrics across supported model choices.
  • Record training and evaluation results in CSV files.
Before you choose

Requirements & limits

The public GitHub repository is identified as open source under the Apache-2.0 license.

  • The repository uses TensorFlow 1.14.0 and specifies Python 3 on Ubuntu 16.04 or 18.04.
  • Task scripts expose model, epoch, hidden-unit size and validation logging options.
  • The documented workflow downloads datasets into a local data folder and saves result CSVs under results.
  • The README points to a sister repository containing a PyTorch version and tutorials.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Train and evaluate a model on a documented sequence task

Prepare a dataset and run its task script to train a selected model, evaluate it, and save result metrics.

Before you start

  • Python 3 and TensorFlow 1.14.0
  • Ubuntu 16.04 or 18.04, as specified in the repository instructions
  • The repository and its dataset download script

Steps

  1. Run source download_datasets.sh to download datasets into the data folder.

  2. Choose a task module: gesture.py, occupancy.py, har.py, traffic.py or ozone.py.

  3. Run the chosen module with a supported --model value (lstm, ctrnn, ltc, ltc_rk or ltc_ex); optionally set --epochs, --size or --log.

  4. Review the validation and test metrics appended to a CSV file in the results folder.

Expected outcome

The script trains the selected model, evaluates validation performance during training, restores its best checkpoint for test evaluation, and records results.

Guide limitations

  • The instructions specify TensorFlow 1.14.0, Python 3 and Ubuntu 16.04 or 18.04.
  • The guide follows the repository instructions; it does not establish execution on other environments.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
raminmh/liquid_time_constant_networks · Official homepage title · retained in its source language
How we review sourcesSuggest a correction