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

Neural Combinatorial Rl Pytorch

Neural Combinatorial Rl Pytorch is a Python/PyTorch research implementation of reinforcement-learning pointer networks for combinatorial optimization, with built-in sorting and planar symmetric Euclidean TSP tasks.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Neural Combinatorial Rl Pytorch — 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 for research experiments with the included sorting or planar symmetric Euclidean TSP tasks, or as a code example for extending pointer-network solution selection to another task.

Who it is for

  • Researchers studying reinforcement-learning approaches to combinatorial optimization.
  • Developers seeking a pointer-network baseline for sorting or planar symmetric Euclidean TSP experiments.
  • Readers inspecting attention in an evaluated saved model.

Documented uses

  • Train a pointer network to construct tours for planar symmetric Euclidean TSP instances.
  • Use the sorting task as a reinforcement-learning baseline.
  • Load a saved model for testing and plot its pointer-network attention.
  • Prototype a new combinatorial task by adding a dataset class and scalar reward function.
Before you choose

Requirements & limits

The implementation is available as a public GitHub repository under the MIT license.

  • The included task scope is sorting and planar symmetric Euclidean TSP.
  • Beam search is unfinished and currently supports only one beam, equivalent to greedy decoding.
  • The README lists Python 3.6, PyTorch 0.2 or 0.3, tqdm, matplotlib, and tensorboard_logger; it notes a separate PyTorch 0.4 compatibility branch.
  • The repository describes adding another task through a dataset class and scalar reward function, but does not include implementations for other tasks.

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

Load a saved model and plot pointer-network attention

Load a saved model for evaluation and request an attention-layer plot.

Before you start

  • The repository code and its documented dependencies: Python 3.6 (the README says versions from 3.4 may work), PyTorch 0.2 or 0.3, tqdm, matplotlib, and tensorboard_logger.
  • A saved model and its path.

Steps

  1. Use the repository's documented main.sh as the example for running the code.

  2. Pass --load_path with the saved model path and set --is_train False.

  3. Also set --plot_attention True to request a plot of the pointer network's attention layer.

Expected outcome

The implementation is configured to load the saved model for testing and plot its attention layer.

Guide limitations

  • The README documents these arguments but does not provide a complete invocation or guarantee a successful run.
  • Test-time decoding is greedy; multi-beam search is unfinished.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
18 October 2026
Original website title
pemami4911/neural-combinatorial-rl-pytorch · Official homepage title · retained in its source language
How we review sourcesSuggest a correction