2.3.6 TF demos

← 2.3 TFs

An LLM consists of:

• Transformer (TF) (the NN)
• Internal agent (iAgent) (interfaces with TF (via tokens) and the agent scripts)

See also _ziptieai_BOOK2_LLM.docx the Gdrive.

Newer demos (2026.0529+)

Demo 5: 2.3.6.1 D5 Tiny TF (26.0529)

Runs locally in CPU mode.

Demo 5b: 2.3.6.1b D5 tiny TF algorithm details (26.0602)

This page describes

• every line of code in the demo
• the details of “attention heads”, FFNs, etc

Demo 11: Tiny TF classifier (TODO)

Demo 12: Tiny TF sentiment detector (TODO)

Demo 13: Tiny TF sequence predictor (TODO)

Older demos (2025)

The wiki page “AI LLM stacks” (a bit chaotic after a recent wiki reorganization) lists the following subpages are of interest:

• 2.2 Demo deployments (HF, CloudFlare, etc)
• 2.3 Youtube demos. I did a lot of YouTube demos. Way back then (2025, a long time ago in AI-time) I had still not make the transition to working totally with LLMs (GPT) to learn new tech.
• 2.4 GPT/Copilot demos (a few demos)
• 2.6 Agent/LLM input docs (RAG demo)

26.0603 (v1 26.0529)