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

Kan Gpt

Kan Gpt is an open-source Python and PyTorch implementation of GPT-style generative language models that uses Kolmogorov-Arnold Networks for language modeling. It includes code for model training, dataset handling and text generation from checkpoints.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Kan Gpt — 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 Kan Gpt when you want to examine or adapt a code-level implementation of GPT-style language modeling with KAN components, or run a documented prompt workflow against a checkpoint.

Who it is for

  • Researchers comparing KAN-based and MLP-based GPT-style language models.
  • Developers prototyping or training text-generation models in Python and PyTorch.

Documented uses

  • Train a GPT-style model using KAN or MLP architecture options.
  • Prepare supported datasets including Tiny Shakespeare and WebText.
  • Generate text from a prompt using a supplied model checkpoint.
Before you choose

Requirements & limits

The repository is public and licensed under MIT. The README provides a PyPI installation command and development setup instructions.

  • This is a model implementation, not a documented hosted inference service or API.
  • The prompt workflow requires a checkpoint; automatic model-weight downloading is listed as unfinished in the README.
  • The project documents both KAN and MLP training paths and a Tiny Shakespeare comparison, but its reported comparison does not establish broader performance.

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

Generate text from a KAN-GPT checkpoint

Use the documented prompt command to generate text from a prompt and an existing checkpoint.

Before you start

  • A Python environment with the project and its dependencies installed.
  • A model checkpoint.
  • The prompt command documented by the repository.

Steps

  1. Install the package with the command below, or clone the repository and install its development dependencies as described in the README.

    Documented command
    pip install kan_gpt

  2. Provide a prompt and checkpoint path to the command below.

    Documented command
    python -m kan_gpt.prompt --prompt "Bangalore is often described as the " --model_path (checkpoint)

Expected outcome

The prompt script produces text using the supplied checkpoint.

Guide limitations

  • The documented command is an example and does not establish a particular output quality.
Documentation 1 ↗
Source-based practical guide

Set up the repository for training

Clone the codebase, download a documented dataset and install its development dependencies.

Before you start

  • Python 3 and pip.
  • Git.
  • A choice of the listed datasets: Tiny Shakespeare, MNIST or WebText.

Steps

  1. Clone https://github.com/AdityaNG/kan-gpt and change into the kan-gpt directory.

  2. Run the dataset download module for the chosen dataset, such as the command below.

    Documented command
    python3 -m kan_gpt.download_dataset --dataset tinyshakespeare

  3. Install the dependencies with the command below.

    Documented command
    pip install -r requirements.txt

  4. Install the local package with the command below.

    Documented command
    pip install -e .

Expected outcome

The repository and selected dataset are prepared according to the documented development setup.

Guide limitations

  • These instructions cover setup only; they do not specify a complete training configuration or guarantee a successful run.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
AdityaNG/kan-gpt · Official homepage title · retained in its source language
How we review sourcesSuggest a correction