↩︎ Home

AI Lab

Two live projects in applied AI for psychoanalytic education

Explore a pilot Virtual Teaching Assistant for dream-seminar learning, plus a Nanochat lab environment for demonstrating how decoding “knobs” change model behavior in real time.

Open Teaching Assistant Open Nanochat Lab

Projects

Open

Virtual Teaching Assistant

A pilot teaching assistant designed for seminars on dream interpretation. Optimized for discussion, synthesis, and learning support.
Launch TA
Login required

Nanochat Lab

Stable and experimental endpoints. Each page displays decoding defaults (temperature, top-k, max tokens) for demonstration and comparison.
Stable Unstable
How to use the Nanochat lab
Use Stable as your baseline. Then open Unstable and try the presets (Stable-ish / Spicy / Chaos) to see how temperature/top-k/max-tokens trade off coherence versus novelty. The best demos come from repeating the same prompt across settings and comparing tone, consistency, and drift.
Need access? Ask Oscar for the guest login.

Enhance Your Learning Experience

Virtual Teaching Assistant
Seminar-first support for learning, discussion, and synthesis.
🧠

Teaching Assistant

A seminar-first assistant designed to support learning and discussion around dream material.

📚

Interactive Seminar Mode

Ask questions, request syntheses, and explore “what-if” readings in a conversational format.

🧭

Scope + Safety

Built for education and demonstration—not clinical decisions, diagnosis, or patient care.

Nanochat Lab — small LLM sandbox
Built from Andrej Karpathy’s Nanochat tutorial codebase; designed for “knob-turning” demos.
🧪

Two Modes (Stable / Unstable)

Same checkpoint, different decoding defaults—so you can compare coherence vs. novelty.

🎛️

Knobs You Can See

Temperature/top-k/max tokens are visible; on Unstable, you can tweak them and watch behavior shift live.

🗂️

What It Trained On

Pretrained on a large sharded text corpus (downloaded by the speedrun scripts), plus small synthetic “identity” chats; later stages add conversation structure and instruction tuning.