zhang0jhon/AttentionOCR

@zhang0jhon tensorflow run in static graph, can you give any hints on debugging during coding?

Opened this issue · 0 comments

It seems that tensorflow is not friendly enough to coders compared to pytorch or mxnet since these two platforms support dynamic graph, so can you give some hints on debugging during coding? I tend to write some extra codes to output the middle results during debugging using tf, however, it is not efficient enough in this way.