Open Yanomizu opened 1 year ago
Hello, I am suffering from exactly the same problem. Did you figure it out? Here is my config (which didn't work):
The problem is with torch1.4.0, I tried to solve it by downgrading and upgrading cuda and cudnn but the problem was the same. So, I upgraded the torch1.4.0 > torch1.7.1 then the problem is solved.
pip install torch==1.7.1+cu101 torchvision==0.8.2+cu101 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html
pip install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cu110/torch_nightly.html -U
Try this, then you should not see the error.
The code may look stuck like below but it'll be loading and processing the data.
To check that you can add for x in tqdm.tqdm(dataloader)
instead of for x in dataloader
on line 50 in train_avd.py
The problem is with torch1.4.0, I tried to solve it by downgrading and upgrading cuda and cudnn but the problem was the same. So, I upgraded the torch1.4.0 > torch1.7.1 then the problem is solved.
pip install torch==1.7.1+cu101 torchvision==0.8.2+cu101 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html
pip install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cu110/torch_nightly.html -U
Try this, then you should not see the error.The code may look stuck like below but it'll be loading and processing the data.
To check that you can add
for x in tqdm.tqdm(dataloader)
instead offor x in dataloader
on line 50 in train_avd.py
Much appreciated! I'll give it a go right away!
The problem is with torch1.4.0, I tried to solve it by downgrading and upgrading cuda and cudnn but the problem was the same. So, I upgraded the torch1.4.0 > torch1.7.1 then the problem is solved.
pip install torch==1.7.1+cu101 torchvision==0.8.2+cu101 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html
pip install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cu110/torch_nightly.html -U
Try this, then you should not see the error. The code may look stuck like below but it'll be loading and processing the data. To check that you can addfor x in tqdm.tqdm(dataloader)
instead offor x in dataloader
on line 50 in train_avd.pyMuch appreciated! I'll give it a go right away!
Just let me know buddy if it is working for you or not, maybe be I must have done few more things that made it work.
The problem is with torch1.4.0, I tried to solve it by downgrading and upgrading cuda and cudnn but the problem was the same. So, I upgraded the torch1.4.0 > torch1.7.1 then the problem is solved.
pip install torch==1.7.1+cu101 torchvision==0.8.2+cu101 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html
pip install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cu110/torch_nightly.html -U
Try this, then you should not see the error. The code may look stuck like below but it'll be loading and processing the data. To check that you can addfor x in tqdm.tqdm(dataloader)
instead offor x in dataloader
on line 50 in train_avd.pyMuch appreciated! I'll give it a go right away!
Just let me know buddy if it is working for you or not, maybe be I must have done few more things that made it work.
Long story short, no, it didn't work out. I ran into another cuda runtime error. But again, I really appreciate your kindness! P.S. I have decided to move on to other works days ago. since I suspect that this bug has something to do with my hardware, I am looking for "fresher" works to work on. XD
The problem is with torch1.4.0, I tried to solve it by downgrading and upgrading cuda and cudnn but the problem was the same. So, I upgraded the torch1.4.0 > torch1.7.1 then the problem is solved. Try this, then you should not see the error. The code may look stuck like below but it'll be loading and processing the data. To check that you can add instead of on line 50 in train_avd.py
pip install torch==1.7.1+cu101 torchvision==0.8.2+cu101 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html``pip install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cu110/torch_nightly.html -U``for x in tqdm.tqdm(dataloader)``for x in dataloader
Much appreciated! I'll give it a go right away!
Just let me know buddy if it is working for you or not, maybe be I must have done few more things that made it work.
Hello, I tried you advice above, but it still can not run. Here are the errors.
Hello.
I am now trying to replicate your research. However, I am having trouble with an error when I try to run the training. I believe the cause is in the version of CUDA, cuDNN. So I would like to know what those are if you were able to run the training. Also, if my guess is wrong, I would like to know the cause.
Thanks.