Sugoi

Installing

Offline

Using CUDA to Improve Performance of Offline Version

NVIDIA GPUs ONLY

Before doing anything, check in NVIDIA control panel > system information > component tab and see what CUDA version you have. v3.xx only works with CUDA v11.xx+, 2.0 works with CUDA 10.xx

Method 1

Using a premade powershell script (Download & read the "install cuda.txt)"

https://cdn.discordapp.com/attachments/795551389211164703/902472195954065428/InstallCudaForSugoiV3.zip

Mega link for those allergic to discord: https://mega.nz/file/k00m2JSa#NmOd2nt6JpIbv1_LOIStQCQwbpS70pMMfUiE6kAn2pg

Method 2

Manually upgrade by pip using python included in Sugoi

1. Open command prompt as administrator (not powershell) go to C:\Sugoi-Japanese-Translator-V3.0\Power-Source\Python38

2. Type:

CUDA 10:

   python.exe -m pip install torch==1.9.1+cu102 torchvision==0.10.1+cu102 torchaudio===0.9.1 -f https://download.pytorch.org/whl/torch_stable.html --user

CUDA 11:

   python.exe -m pip install torch==1.9.1+cu111 torchvision==0.10.1+cu111 torchaudio===0.9.1 -f https://download.pytorch.org/whl/torch_stable.html --user

3. In backendServer\Program-Backend\Sugoi-Translator-Offline\offlineTranslation\fairseq\flaskServer.py change line 19 # ja2en.cuda() to ja2en.cuda()

Method 3

Manually upgrade by pip using global python installation on Windows

1. Open command prompt (or powershell) as administrator and type:

CUDA 10:

   pip install --target=C:\Sugoi-Japanese-Translator-V3.0\Power-Source\Python38\Lib\site-packages --upgrade torch==1.9.1+cu102 torchvision==0.10.1+cu102 torchaudio===0.9.1 -f https://download.pytorch.org/whl/torch_stable.html

CUDA 11:

   pip install --target=C:\Sugoi-Japanese-Translator-V3.0\Power-Source\Python38\Lib\site-packages --upgrade torch==1.9.1+cu111 torchvision==0.10.1+cu111 torchaudio===0.9.1 -f https://download.pytorch.org/whl/torch_stable.html

2. Change line 19 in backendServer\Program-Backend\Sugoi-Translator-Offline\offlineTranslation\fairseq\flaskServer.py from # ja2en.cuda() to ja2en.cuda()