Then you're going to have to inspect the file for things that look like likely floating-point values since there are no specific values that you can search for in the file. You could try posting a small initial portion of the data here – 256 bytes or so.
On Wed, Oct 26, 2016 at 5:24 PM, Aleksandr Mikheev <al.mikhee...@gmail.com> wrote: > Unfortunately, I do not. However, I am most definitely sure that > everything is okay with file since it is a training file. > > среда, 26 октября 2016 г., 23:17:00 UTC+3 пользователь Stefan Karpinski > написал: >> >> Do you have an example where you have the file and you know what the data >> should be? You may want to look at the data in a hex editor and try to >> figure out what's going on. >> >> On Wed, Oct 26, 2016 at 4:06 PM, Aleksandr Mikheev <al.mik...@gmail.com> >> wrote: >> >>> So, to avoid the opening a new thread I will write here again. It seems >>> I still have a problem which I cannot solve. I use >>> >>> a_million_floats = read(f, Float64, 1_000_000) >>> >>> >>> as it was suggested, however the numbers I get are obviously wrong >>> (there are numbers like 8.68e+272, -5.39e-142, and even NaN). I am pretty >>> that these are not the right numbers. Also, I am pretty sure that the type >>> is correct (it should be 'double' in C/C++). Furthermore, I have i5 CPU, so >>> it should be little endian as it is supposed in .dat file. What could be >>> wrong? >>> >>> Thanks in advance! >>> >> >>