EMHussain

error :``` SIGSEGV Segmentation violation (invalid memory reference) /data/app/ch.futureit.suchagentnative--7UzNnHYX1Um0ywhwcNdww==/lib/arm64/libmyhtml.so:102776 mycore_string_copy``` example html [body-load-01.log](https://github.com/lexborisov/myhtml/files/6316909/body-load-01.log)

Hey, have you solve this question？ or can you give me some suggestion about it？ Thanks，best wish

I have tried `torchscript model` on python got same output as in C++ ``` model = jit.load("parseq.torchscript") model.eval() img_transform = SceneTextDataModule.get_transform((32, 128)) image = Image.open('cropped.jpeg').convert('RGB') image = img_transform(image).unsqueeze(0).to('cpu') p =...

I was doing mistake here `p.softmax(-1)` it should be `p = p.softmax(-1)` and in c++ it should be `output = output.softmax(-1);`

After trying different version of Pytorch in C++ got correct output on torch version 1.13.1 but still issue in inference time that is almost 7 times high in C++ as...