(function(){ try { // Neon-leaning palette for vivid output var hues = [50, 90, 120, 140, 155, 170, 185, 200, 215, 235, 255, 270, 285, 300, 315, 330, 345]; var H = hues[Math.floor(Math.random() * hues.length)]; var root = document.documentElement; var isDark = root.classList.contains('dark'); function setVar(name, h, s, l){ root.style.setProperty(name, h+" "+s+"% "+l+"%"); } function srgbToLinear(c){ return c <= 0.04045 ? c/12.92 : Math.pow((c+0.055)/1.055, 2.4); } function hslToRgb(h,s,l){ var C=(1-Math.abs(2*l-1))*s, Hp=h/60, X=C*(1-Math.abs((Hp%2)-1)); var r=0,g=0,b=0; if(0<=Hp&&Hp<1){r=C;g=X;} else if(1<=Hp&&Hp<2){r=X;g=C;} else if(2<=Hp&&Hp<3){g=C;b=X;} else if(3<=Hp&&Hp<4){g=X;b=C;} else if(4<=Hp&&Hp<5){r=X;b=C;} else if(5<=Hp&&Hp<6){r=C;b=X;} var m=l-C/2; return [r+m,g+m,b+m]; } function relativeLuminance(h, sPct, lPct){ var s=sPct/100, l=lPct/100; var rgb=hslToRgb(H, s, l); var R=srgbToLinear(rgb[0]), G=srgbToLinear(rgb[1]), B=srgbToLinear(rgb[2]); return 0.2126*R + 0.7152*G + 0.0722*B; } function ensureContrast(h, sPct, initialLPct, darkMode){ var minRatio=4.5; var targetYAgainstWhite=(1.0+0.05)/minRatio - 0.05; var targetYAgainstBlack=minRatio*0.05 - 0.05; var low=0, high=100, l=initialLPct; for(var i=0;i<14;i++){ var y=relativeLuminance(h, sPct, l); if(!darkMode){ if(y <= targetYAgainstWhite){ low=l; l=(l+high)/2; } else { high=l; l=(l+low)/2; } } else { if(y >= targetYAgainstBlack){ high=l; l=(l+low)/2; } else { low=l; l=(l+high)/2; } } } return Math.max(0, Math.min(100, l)); } var s=100; var baseL=ensureContrast(H, s, isDark ? 66 : 50, isDark); var lMain=isDark ? Math.max(baseL, 64) : baseL; setVar('--accent', H, s, lMain); var l2=Math.max(0, lMain - (isDark ? 8 : 12)); var l3=Math.min(100, lMain + (isDark ? 12 : 10)); var l4=Math.min(100, lMain + (isDark ? 24 : 18)); setVar('--accent2', H, s, l2); setVar('--accent3', H, s, l3); setVar('--accent4', H, s, l4); var y=relativeLuminance(H, s, lMain); var accentFg = y > 0.5 ? '0 0% 0%' : '0 0% 100%'; root.style.setProperty('--accent-foreground', accentFg); } catch(_){ } })();

Investigating Demographic Attribute Representation in Vision Encoders