Handbook
ASPIRE stands for Adversarial Student-Professor Internalized Reasoning Engine. The core idea: teaching AI judgment through internalized mentorship, not just memorizing answers.
Traditional fine-tuning says “here are the right answers, match them.” ASPIRE says “here is a wise mind, learn to think like it does.” When you learn from a great mentor you don’t memorize their answers. You internalize their way of seeing. Their voice becomes part of your inner dialogue. You start to anticipate what they would say, and eventually that anticipation becomes your own discernment.
ASPIRE gives AI that same experience.
What you will find here
Section titled “What you will find here”- Beginners Guide — New to ASPIRE? Start here for a plain-language walkthrough.
- Getting Started — Install, configure, and run your first dialogue.
- How It Works — The four-stage pipeline from adversarial dialogue to judging responses without the teacher.
- Teachers — The five teacher personas and how to compose them into committees.
- Integrations — Stable Diffusion Forge, Isaac Gym, and code assistant integrations.
- CLI Reference — Every command, every flag, every option.
- Watching Runs in ScalarScope — Export a run’s training dynamics and compare runs side by side.
ASPIRE also includes an experimental perception module with theory of mind, metacognition, character persistence, and controlled chaos capabilities for building agents with deeper awareness.
The short version
Section titled “The short version”A student model generates responses. A teacher model challenges them through adversarial dialogue. A critic model learns to predict the teacher’s judgment. The student then trains against the critic’s internalized feedback. After training, the critic judges the student’s responses alone — no teacher API calls needed.