Soumith Chintala
Soumith Chintala
I have no energy to write this. if there's an incoming PR that implements it, I'll happily accept it.
i am not sure either. but i have no interest investigating, considering that inception-v3.torch exists: https://github.com/Moodstocks/inception-v3.torch
I am not able to reproduce the issue. Can you try audio.save with a different output format, such as test_out.ogg or test_out.wav
Does this issue reproduce with audio.samplevoice() or is it just with your specific file?
yes, i dont do that. kinda surprising that no one else noticed. However, because of precision limits, symmetry will slowly break..., after an epoch or so.
same with ssh, does not work.
oh, it seems to be an issue with screen. I tried it in ssh, without screen, and everything works.
same with imgcat actually, that also does not work inside screen.
yea i saw that, but that's working only for tmux, looks like it. i'll investigate into what's needed for screen.
elseif src.__nDimension == 1 and src.__size[0] == 1 then return src:data()[0] ... I am keeping this issue for tracking bugs, you dont need to do anything for now.