Problem in fastai with Nvidia Titan 1080


I tried running

interp = ClassificationInterpretation.from_learner(learn)
and getting Floating point exception (core dumped) and the process stops

What can be possible causes for that?

similarly i am getting error now in
learn.fit_one_cycle(1, 1e-2, moms=(0.8,0.7))

after epoch the process stop with same floating point exception

What could be the bug? nvidia driver or fastai code

Hi Yash,

  1. It sounds like you are trying to run in mixed-precision with Fp32/Fp16, correct ?

  2. Do you have a Titan or a 1080 gpu ? That can be a very different architecture.

Hi Eric,

  1. I don’t know if I am doing that, I am just running fastai course notebooks on gpu and got this problem. How mixed-precison is there I have no idea in that regards. Will need your help
  2. I have titan xp