Every hash table implementation I have done I use an RB tree for each bucket. This way if the bucket is loaded with hundreds or thousands of elements, you aren’t penalized too badly.