Language Model Response Laboratory¶
Run the Language Model Response Laboratory MicroSim Fullscreen
Edit in the p5.js Editor
About This MicroSim¶
This worked example makes four otherwise abstract ideas visible: instructional roles, simplified tokens, finite context, and illustrative next-phrase probabilities. It explicitly distinguishes stable meaning from variable wording without claiming access to proprietary model internals.
How to Use¶
Make a prediction, then use Next and Previous to inspect each stage. At the context stage, add irrelevant material; at the probability stage, vary the illustrative distribution and compare the two final responses.
Iframe Embed Code¶
You can add this MicroSim to any web page by adding this to your HTML:
<iframe src="https://dmccreary.github.io/ai-persona-testing/sims/language-model-response-lab/main.html"
height="717"
width="100%"
scrolling="no"></iframe>
Lesson Plan¶
Grade Level¶
High school, undergraduate, and professional AI literacy
Duration¶
10-15 minutes
Prerequisites¶
No programming experience is required; learners should understand that language models generate text from prompts.
Activities¶
- Predict (3 min): Choose which input will most influence the response.
- Step through (7 min): Explain what changes at each processing stage.
- Compare (5 min): Identify stable meaning and variable wording across two runs.
Assessment¶
Learners can explain the role of instructions, tokenization, context limits, and probabilistic selection without treating the teaching probabilities as actual internal model values.
References¶
- Attention Is All You Need - Foundational transformer architecture paper.
- p5.js Reference - Documentation for the interactive rendering library.