accumulate_grad warning when train convnext_tiny with pytorch_lightning and accumulate #1104
              
                Unanswered
              
          
                  
                    
                      ray-lee-94
                    
                  
                
                  asked this question in
                Q&A
              
            Replies: 1 comment 2 replies
-
| @VCBE123 you'll have to provide a lot more detail of your environment, hardware, model etc. I do have an idea what could cause it, but I can't reproduce on PyTorch 1.10+ on 3090 and there were some incorrect triggers for the warning in older versions of PT. It is also as it says, a warning and may not be impacting anything. | 
Beta Was this translation helpful? Give feedback.
                  
                    2 replies
                  
                
            
  
    Sign up for free
    to join this conversation on GitHub.
    Already have an account?
    Sign in to comment
  
        
    
Uh oh!
There was an error while loading. Please reload this page.
-
[W accumulate_grad.h:170] Warning: grad and param do not obey the gradient layout contract. This is not an error, but may impair performance.
grad.sizes() = [96, 1, 7, 7], strides() = [49, 49, 7, 1]
param.sizes() = [96, 1, 7, 7], strides() = [49, 1, 7, 1] (function operator())
Beta Was this translation helpful? Give feedback.
All reactions