Hello, in the original version of the exam we had "f(x) < tau" and not "<=".
I know that, at some point, someone said that it was a typo and the correct version was "<=", but still I personally don't think it's really fair to change the question during the exam (and not even that early).
In my case, I had already solved this problem with the original text (and marked the first answer on the published version, as, in this case, the output of h with tau=0 was always NO) and I was far ahead when we were told about the typo. Of course, I gave it a look after that, but I didn't remember exactly the details of the question and so I did not manage to change my answer. The exam was already quite long and, personally, I don't think it's really fair to make us solve the same problem again because of a typo...
I'm sorry to bring this up, but I just wanted to know if it was possible to accept also the answer to the original text of the exam.
Thanks in advance.
Personally, I did not answer the original question with the typo because, from my understanding, the 4th answer (with q) would be just as correct as the first one so there was no good answer. When the typo was announced I came back to the question (I'm not sure why the teacher hinted that it was no big deal, though)
For me, I spend some time on the question and saw that the first answer on the published version was ok. When the typo was announced, I erased my answer but didn't have the time to re-think the question.
@Anonymous said:
Hello, in the original version of the exam we had "f(x) < tau" and not "<=".
I know that, at some point, someone said that it was a typo and the correct version was "<=", but still I personally don't think it's really fair to change the question during the exam (and not even that early).
In my case, I had already solved this problem with the original text (and marked the first answer on the published version, as, in this case, the output of h with tau=0 was always NO) and I was far ahead when we were told about the typo. Of course, I gave it a look after that, but I didn't remember exactly the details of the question and so I did not manage to change my answer. The exam was already quite long and, personally, I don't think it's really fair to make us solve the same problem again because of a typo...
I'm sorry to bring this up, but I just wanted to know if it was possible to accept also the answer to the original text of the exam.
Thanks in advance.
if you use < tau then none of the choices are correct. its true that the relu methods are the same, the sigmoid methods are the same, but sigmoids will work as normal but relu ones will always predict the same value. so g and f are the same, h and q are the same, but g and f != h and q.
Exam 2021 - Q9 - Typo in the exam
Hello, in the original version of the exam we had "f(x) < tau" and not "<=".
I know that, at some point, someone said that it was a typo and the correct version was "<=", but still I personally don't think it's really fair to change the question during the exam (and not even that early).
In my case, I had already solved this problem with the original text (and marked the first answer on the published version, as, in this case, the output of h with tau=0 was always NO) and I was far ahead when we were told about the typo. Of course, I gave it a look after that, but I didn't remember exactly the details of the question and so I did not manage to change my answer. The exam was already quite long and, personally, I don't think it's really fair to make us solve the same problem again because of a typo...
I'm sorry to bring this up, but I just wanted to know if it was possible to accept also the answer to the original text of the exam.
Thanks in advance.
2
Personally, I did not answer the original question with the typo because, from my understanding, the 4th answer (with q) would be just as correct as the first one so there was no good answer. When the typo was announced I came back to the question (I'm not sure why the teacher hinted that it was no big deal, though)
For me, I spend some time on the question and saw that the first answer on the published version was ok. When the typo was announced, I erased my answer but didn't have the time to re-think the question.
1
if you use < tau then none of the choices are correct. its true that the relu methods are the same, the sigmoid methods are the same, but sigmoids will work as normal but relu ones will always predict the same value. so g and f are the same, h and q are the same, but g and f != h and q.
1
Add comment