Hi,
I believe there's an issue with the 5th test case in human_eval problem_103. Currently it is set to have 366 as result
-- #test implementation 185 546 = some "0b101101110"
however upon uncommenting the test fails for the provided implementation.
The result of the computation is 365.5, which should be rounded down to 365 in accordance to the implementation, the other test cases as well as the provided test cases in the NL description.
Note that the NL docstring is however unclear on how to handle non-integer results: it states "Round the answer to the nearest integer" which conventionally means half-up rounding on .5, I believe.
I will submit a PR addressing this shortly.
Hi,
I believe there's an issue with the 5th test case in human_eval problem_103. Currently it is set to have 366 as result
-- #test implementation 185 546 = some "0b101101110"however upon uncommenting the test fails for the provided implementation.
The result of the computation is 365.5, which should be rounded down to 365 in accordance to the implementation, the other test cases as well as the provided test cases in the NL description.
Note that the NL docstring is however unclear on how to handle non-integer results: it states "Round the answer to the nearest integer" which conventionally means half-up rounding on .5, I believe.
I will submit a PR addressing this shortly.