A simple way to implement is to compare every complete path with the the previous paths, which actually works, but not good enough.

One way to improve is to record the previous value of candidate, if the new candidates's values equal to the recorded one, continue.

Pay attention to the position that the variable 'prev' is declared. It can make sure that the for loop won't continue at the first place.

```
def helper(result, path, nums, m, sum, target):
if sum == target:
#if path not in result: the simple way to implement, need to do search everytime
result.append(path[:])
return
prev = 0
for i in range(m, len(nums)):
if target - sum < nums[i]:
return
if nums[i] == prev: #record the previous visited candidates, can avoid duplicates at the first place
continue
path.append(nums[i])
prev = nums[i]
helper(result, path, nums, i + 1, sum + nums[i], target)
path.pop()
result = []
path = []
candidates.sort() # put the duplicate candidates together
s = 0
m = 0
helper(result, path, candidates, m, s, target)
return result
```