One approach is to use itertools.chain to glue sublists together
>>> list(itertools.chain(*[[k]*v for k, v in d.items()]))
[1, 1, 1, 10, 10, 5, 5, 5, 5, 5, 5]
Or if you are dealing with a very large dictionary, then you could avoid constructing the sub lists with itertools.chain.from_iterable and itertools.repeat
>>> list(itertools.chain.from_iterable(itertools.repeat(k, v) for k, v in d.items()))
[1, 1, 1, 10, 10, 5, 5, 5, 5, 5, 5]
Comparative timings for a very large dictionary with using a list comprehension that uses two loops:
>>> d = {i: i for i in range(100)}
>>> %timeit list(itertools.chain.from_iterable(itertools.repeat(k, v) for k, v in d.items()))
10000 loops, best of 3: 55.6 µs per loop
>>> %timeit [k for k, v in d.items() for _ in range(v)]
10000 loops, best of 3: 119 µs per loop
It's not clear whether you want your output sorted (your example code does not sort it), but if so simply presort d.items()
# same as previous examples, but we sort d.items()
list(itertools.chain(*[[k]*v for k, v in sorted(d.items())]))