C
Concepts
Nineteen ML and search explainers you learn by dragging the inputs around
Emerged finds new products worth a look, daily — what’s new today
LLM Attention Visualization
Hover any generated token to see which earlier tokens shaped it
It makes attention visible in a way that finally explains why LLMs copy addresses, dates and code so reliably.
An interactive in-browser explainer running a small 600M-parameter model. Tap or hover over any generated token and the earlier tokens that influenced it light up, weighted by attention scaled by value-vector magnitude and summed across heads and layers. Preset prompts show verbatim copying and code edits clearly. The author is upfront that the single-number view is a simplification.
More like this
HN.watch
The Hacker News front page, each story turned into a short narrated explainer
anything2explainer
A Claude Code skill that turns any topic into a narrated explainer video
glassbrain
A browser viewer and diff tool for the knowledge graphs your agents write
RelateAnything
Open-vocabulary relation prediction between any image regions, in real time
VisualFries
An MIT render engine your coding agent drives to cut raw footage into posts