Open akshowhini opened 4 years ago
calling the function triggers to compute the same output
In the training mode,an error occurred ' CUDA out of memory',in the first epoch,needing 6.63G,but in the second epoch,CUDA needs 7G,how can i solve this problem?i think that forward let out.
calling the function triggers to compute the same output