Yupan Huang

Thanks for your contribution. Based on @mtanti 's implementation, I modified two places to support meteor evalution for both py2 and py3. 1. It seems that the code of ```...

Thanks @mtanti for pointing it out! I've modified the code. @ozancaglayan Since I use the code of some repositories which only support python2 originally, I am transferring to python3 and...

Hi @johnnyjana730 , I am struggling with building up the env too. Could you kindly share me your docker env? Since I cannot find your email, could you contact me?...

Hi @fengyang0317 , thanks for your contribution! How many GPUs do you use for your default setting? To train with initialization, I use your suggest command as follows, ``` python...

Hi @fengyang0317 , I tried the multi-gpu and multi-thread setting for evaluation. It seems that the time cost is the same as manually evaluate each model with different gpus in...

Thanks and looking forward to your related files of unpaired setting!

I met the similar problem whose output error saying `./make.sh nvcc command not found`. I solved it by ``` export LD_LIBRARY_PATH=/usr/local/cuda/lib export PATH=$PATH:/usr/local/cuda/bin ``` and make again. Reference: https://devtalk.nvidia.com/default/topic/457664/nvcc-quot-no-command-nvcc-found-quot-/

Hi, There's a little problem to run with the provided command in [this README](https://github.com/rowanz/r2c/tree/master/models) like `python train.py -params multiatt/default.json -folder saves/flagship_answer` in the current directory structure. Simple modification can work:...

Please see section 3.3 Fine-tuning on Multimodal Tasks, Task III Document Visual Question Answering in the paper.

Thanks @jordanparker6 for the replies. For the Cuda/Pytorch-related problem, perhaps @rusubbiz-muzkaq can refer to other related answers and reinstall PyTorch with a proper version of Cuda.