So everybody please write what you have. Using a smaller
subset is good thing to try.
We will discuss today about the difficulties of this hw.
※ 引述《ansel2025 (Ansel)》之铭言：
: ※ 引述《mercurial (多变)》之铭言：
: : I estimate that it will need about 30 hours to run 1nn for 520000 and
: : covtype.test.
: : So I give up to use the whole data to calculate.
: : I only use 1/10 of the data to run 1nn, and it costs me 10 mins.
: I need 20 hours to come out the answer. I run it from last night. If
: I am lucky, maybe I can get the answer tonight. But maybe tomorrow morning
: or longer. It's like gamble.