In　this　very　selective　overview,　we　summarise　the　recent　developments　by　our　own　and　other,　on　the　empirical　likelihood　in　some　nonparametric　and　semiparametric　regression　models.　The　models　include　the　partially　linear　model,　the　single-index　model,　the　partially　linear　single-index　model,　the　varying　coefficient　model,　and　so　on.　The　focus　of　this　overview　is　to　expatiate　the　adjustment　and　＂bias　correction＂　methodologies　when　Wilks＇　phenomenon　does　not　hold.　The　adjustment　or　bias　correction　can　make　the　limiting　distributions　tractable　such　that　they　can　be　directly　used　to　construct　the　confidence　regions　of　parameters　of　interest　without　the　assistance　of　Monte　Carlo　approximation.