# Meaning Machine — A Visual Explorer for How LLMs Simulate Understanding

**URL:** <https://discuss.huggingface.co/t/meaning-machine-a-visual-explorer-for-how-llms-simulate-understanding/152266>\
**Category:** Show and Tell\
**Created:** [April 25, 2025, 8:41am UTC](https://discuss.huggingface.co/t/meaning-machine-a-visual-explorer-for-how-llms-simulate-understanding/152266 "2025-04-25T08:41:18Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![jdspiral](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/jdspiral/32/46130_2.png) [@jdspiral](https://discuss.huggingface.co/u/jdspiral)\
**Post date:** [April 25, 2025, 8:41am UTC](https://discuss.huggingface.co/t/meaning-machine-a-visual-explorer-for-how-llms-simulate-understanding/152266/1 "2025-04-25T08:41:18Z")

</div>

Hey everyone — I just finished building [**Meaning Machine**](https://meaning-machine.streamlit.app/), a visual tool that lets you see what happens when you input a sentence into a language model.

- Tokenization with BERT (via Hugging Face 🤗)
- POS tagging and dependency parsing with spaCy
- Embedding space visualized with PCA + Plotly
- SVO extraction for structure and meaning

It’s meant to show how models “understand” text statistically — not with real-world grounding, but through structure, frequency, and co-occurrence.

Would love feedback on the concept, UX, or how useful this might be for teaching/explainability.

Live: [https://meaning-machine.streamlit.app](https://meaning-machine.streamlit.app)  
Context: [Parsing Perception - by Joshua Hathcock](https://theperformanceage.com/p/how-language-models-see-you)  
GitHub: [GitHub - jdspiral/tokenizer](https://github.com/jdspiral/meaning-machine)