Transformer-OCR

Transformer-OCR copied to clipboard

@delveintodetail Is there wrong in the predict.py file?

2

@delveintodetail Is there wrong in the predict.py file? _Originally posted by @li10141110 in https://github.com/fengxinjie/Transformer-OCR/issues/4#issuecomment-607083558_ for epoch in range(10000): model.train() run_epoch(train_dataloader, model, SimpleLossCompute(model.generator, criterion, model_opt)) model.eval() test_loss = run_epoch(val_dataloader, model, SimpleLossCompute(model.generator,...

delveintodetail

Really interesting technique, good work! Could you please put a License file, so that I can use it in my project? Thanks in advance!

yaysummeriscoming

RuntimeError: cuda runtime error (59) : device-side assert triggered at /opt/conda/conda-bld/pytorch_1565272279342/work/aten/src/THC/generic/THCTensorScatterGather.cu:380

4

请问这个是什么原因，和cuda版本有关系吗？

wyc2015fq

Hi, I notice you have a heatmap visualization of the source attention. I also tried to visualize it by referring to the link http://nlp.seas.harvard.edu/2018/04/03/attention.html but no success so far. Could...

hiepph

You should carefully read the SVTP images.

3

SVTP has 645 images, if you carefully read each one, you will find there are around 20 images humans cannot recognize. How can this algorithm get 98.6%?? When seeing such...

delveintodetail