Comment by kveykva

Is there a specific explanation about how this is better or different than vega itself? https://vega.github.io/vega/docs/specification/

My understanding is that Vega was already an expressive DSL for visualizations and its probably already well spread through LLM training data.