Skip to content

jev-grug

Toy chat that makes Jev talk by repeatedly choosing its next word from a tiny caveman vocabulary.

Projectprototype

Overview

grug.chat is a playful experiment that gets Jev to hold a conversation even though it does not generate text. It repeatedly asks a Choice question for the next word in Grug's reply, from a compact vocabulary plus an 'end' option, using the conversation and words so far as state, until Jev picks end. Grouped tournaments narrow the vocabulary for each word.

The author notes this is a deliberate inversion of Jev's intended use, not a recommended chatbot design. A live demo is available; locally it calls the real API when TYPESAFE_API_KEY is set and otherwise falls back to a deterministic mock so contributors can run the interface.

What the decision model decides

TypeQuestionAnswers
ChoiceChoose the next word in Grug's reply (or end)—

How it works

  1. Take the conversation and words so far as state
  2. Run a grouped Choice tournament over the vocabulary
  3. Append the chosen word or stop on end
  4. Repeat

Strengths

  • Fun illustration of Choice
  • Mock mode without a key

Caveats

  • Toy; not a production chatbot design
  • Many calls per reply

Getting started

Try the live demo at https://jev-grug-chat.mkotlikov.chatgpt.site/ or run locally with TYPESAFE_API_KEY set.

At a glance

Type
Project
Status
prototype
Language
TypeScript
Runtime
Web app
Has code
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

How does Grug talk?

Jev chooses each next word from a small vocabulary.

Is this how Jev should be used?

No, the author calls it a toy inversion of its intended use.

Can I run it without a key?

Yes, it falls back to a deterministic mock.

Related projects

  • Local browser app that turns Jev's per-pixel colour probability distributions into textured paintings.

    Browser extensionprototypeJavaScript
  • Text generation with Jev via cascaded Choice questions: word type, WordNet category, embedding cluster, then the word itself.

    ProjectresearchPython
  • Experiment that uses Jev as a sampler: DeepSeek proposes top-k next tokens and Jev chooses which comes next and when to stop.

    ProjectresearchPython
  • Experiment that makes Jev produce prose by asking one choice question per word over a routed domain vocabulary.

    LibraryresearchJavaScript
  • Playful web arena where Jev does arithmetic, writes chat replies and draws pixel art one Choice or Noul decision at a time.

    Demoprototype
  • ChatGPT-style web chat where every word of every reply is a Jev multiple-choice decision over a ~1,800-word vocabulary plus END.

    ProjectprototypeTypeScript