Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

jev-youtube-demo

Video demo classifying 50 mock YouTube comments with Jev (intent, value, video idea) and comparing speed and cost with GPT-4o-mini.

Modelprototype
View on GitHub ↗

Overview

Jev YouTube Comment Analyser is the demo for a CodeTutor video on judgement models. It classifies 50 mock YouTube comments with Jev, asking three questions per comment in a single parallel call: the intent as a Choice among praise, question, content request, criticism and spam; the audience insight value as a Score from 1 to 5; and whether the comment contains a future video idea as a Noul probability. Up to ten requests run in parallel, and high-value comments with likely ideas are starred.

A second script runs the exact same 50 comments and three judgments through GPT-4o-mini for a direct speed and cost comparison. It is aimed at creators and developers learning how typed judgments differ from prompting an LLM. Setup needs a TypeSafe key (or access via Vercel AI Gateway or OpenRouter) and an OpenAI key for the comparison.

What the decision model decides

TypeQuestionAnswers
ChoiceWhat is this comment's intent?praise · question · content_request · criticism · spam
ScoreHow much audience insight value does this comment have (1-5)?1 · 2 · 3 · 4 · 5
Yes / noDoes this comment contain a future video idea?—

How it works

  1. Load 50 mock comments
  2. Ask three questions per comment in one call, ten in parallel
  3. Print intent, value and idea probability; star ideas
  4. Run the same set through GPT-4o-mini for comparison

Strengths

  • Simple side-by-side speed and cost comparison
  • Shows all three primitives in one request

Caveats

  • Mock comments only
  • Comparison results not reported in the README

Getting started

pip install -r requirements.txt, add TYPESAFE_API_KEY and OPENAI_API_KEY to .env, then `python demo.py` and `python demo_openai.py`.

At a glance

Type
Model
Status
prototype
Language
Python
Runtime
CLI
Has code
yes
Has tests
no

Use cases and integrations

How curated lists describe it

Frequently asked questions

What does Jev judge for each comment?

Intent, insight value from 1 to 5, and the probability it contains a video idea.

What is the comparison?

The same comments and questions run through GPT-4o-mini for speed and cost.

Are the comments real?

No, they are 50 mock comments.

Related projects

  • taiwan-sovereignty-decision-models

    Study finding commercial decision models including Jev give different yes/no probabilities on Taiwan statements by Chinese script.

    ServiceresearchPython
  • jev-music-theory-1

    Experiment testing whether Jev knows music theory via SATB part-writing and quizzes vs Claude, plus a Jev-driven Rhodes piano demo.

    AppresearchJavaScript
  • jev-classify-knn-comparison

    Benchmark of zero-shot Jev versus a PyTorch MLP and k-NN on embeddings for Jigsaw toxic comment classification.

    BenchmarkresearchPython
  • ComfyUI-SystemOne

    ComfyUI nodes that branch workflows on typed Choice, Yes/No and Score judgments from Laya, Jev or Cloudflare Clef.

    ServicebetaPython
  • social-risk-qwen-jev

    Chinese multimodal moderation research pairing a Qwen3-VL LoRA evidence extractor with Jev policy decisions, benchmarked against baselines.

    ModelresearchPython
  • jevcut

    Auto-clipper for talky videos: code enumerates every possible cut point and Jev judges where each standalone clip starts and ends.

    BenchmarkbetaPython