arxiv: https://arxiv.org/abs/2106.11539 this work proposes a backbone for visual document understanding domain. It uses text, visual, spatial features. Key points use text, visual, spatial features at each encoding layer, keep feeding in visual and spatial features on the input side. This has the ‘residual’ connection effect. text and visual features Read more…
arxiv: https://arxiv.org/abs/2012.14740 LayoutLMv2 is a work improved from LayoutLM. This line of work focues on creating a model for document understanding. The main task of this genre is to extract useful information when a document is given. In this case, it tries to extract information when a document image is Read more…
Background I installed pycuda and was trying to run a file and error like this occured Solution After googling, I found this post saying that this was due to lower version of libstdc++.so.6 originally my /lib64/libstdc++.so.6 was pointing to /lib64/libstdc++.so.6.0.19. But the post suggested to use a version higher. First, Read more…
Background I installed pycuda with pip install pycuda==2020.01 and it installed without any errors. However when running the code, I got the following error. appartenly importing pycuda itself was causing error. Fix I removed the install pycuda. downloaded pycuda source files(version 2020.01) and build&installed it from source. After doing this, Read more…
Background I downloaded another CUDA toolkit(11.4) installation .sh and executed it to install it. However, while installing it was stopped due to no disk space. The default install location is /usr/local/cuda-11.4 and unfortunately I only had about 5GB left under / while I had a lot of space under /data. Read more…
Background I was trying to convert pytorch bert-like model from torch to onnx and see if I can run it in onnxruntime. Here’s the environment I used python: 3.9 torch: 1.9.0 onnx: 1.10.2 onnxruntime-gpu: 1.9.0 I converted by pytorch model to onnx with the following line then I tried to Read more…
arxiv: https://arxiv.org/abs/2108.04539 key points use text and spatial information. doesn’t utilize image feature a better spatial information encoding method compared to LayoutLM propose new pretraining task: Area Masked Language Model spatial information encoding method For each text box, get four corner point x,y coordinates and normalize them all with image Read more…