3 ms·
No experience actually trying, but I'd suspect the first step would be training a system on combinations of CSS and HTML; all the really interesting behavior co
by azernik 11y ago
No experience actually trying, but I'd suspect the first step would be training a system on combinations of CSS and HTML; all the really interesting behavior comes from the interaction between the two.
Then I imagine you could do interesting stuff by e.g. constraining the HTML and seeing what kind of CSS was spit out.