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

Value Iteration Networks

Value Iteration Networks is a Python and TensorFlow implementation of the Value Iteration Networks paper, with code for training on GridWorld datasets and an 8×8 dataset included in the repository.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Value Iteration 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 to study or reproduce the repository’s documented GridWorld training workflow, or as a code reference when adapting experiments to related grid-based planning tasks.

Who it is for

  • Researchers and students studying Value Iteration Networks or reinforcement-learning models.
  • Developers seeking a TensorFlow reference implementation for GridWorld planning experiments.

Documented uses

  • Train the model on the included 8×8 GridWorld dataset.
  • Inspect a TensorFlow implementation based on the paper and the authors’ earlier Theano code.
  • Explore training-progress monitoring with the documented TensorBoard logging setup.
Before you choose

Requirements & limits

The source is a public GitHub repository listed under the Apache-2.0 license.

  • The documented default is the included 8×8 GridWorld dataset; 16×16 and 28×28 datasets are linked from the original authors’ repository.
  • The README reports about 98.5% accuracy in under 30 epochs for 8×8, compared with about 99.6% listed in the paper.
  • The repository documents Python 3.6+, TensorFlow 1.0+, and SciPy 0.18.1+ for data loading.
  • Other domains are discussed in reference to the original model; this repository’s described training workflow is GridWorld.

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 the included 8×8 GridWorld model

Run the repository’s documented default training setup on its included 8×8 GridWorld dataset.

Before you start

  • Python 3.6 or later
  • TensorFlow 1.0 or later
  • SciPy 0.18.1 or later to load data
  • The repository and its included 8×8 GridWorld dataset

Steps

  1. Install the documented dependencies in a suitable Python environment.

  2. From the repository, run the command below to start the default 8×8 GridWorld training.

    Documented command
    python3 train.py

  3. To monitor progress, set config.log to True and launch tensorboard --logdir /tmp/vintf/; change config.logdir if using another log directory.

Expected outcome

The documented training process runs the 8×8 GridWorld model; the README reports convergence in under 30 epochs at about 98.5% accuracy.

Guide limitations

  • The reported accuracy and convergence are README results, not a guarantee for other setups.
  • The README documents 8×8 GridWorld as the default; it does not provide setup steps here for the 16×16 and 28×28 datasets.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
18 October 2026
Original website title
hi-abhi/tensorflow-value-iteration-networks · Official homepage title · retained in its source language
How we review sourcesSuggest a correction

Related options

The authors’ original Theano implementation — The README identifies it as the original implementation on which this TensorFlow code is based.