Loss NAN

[itbear-zane]

HI, there. Thank you for your contribution. After training for a few epochs (using the training example cabin, no more changes), loss = NaN ocurred. loss=nan (nan), lr=0.000100: : 38% 38/100 [01:47<03:04, 2.97s/it]WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor. WARNING:tensorboardX.x2num:NaN or Inf found in input tensor..