New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Why different Softmax? #8

Open

LukeAI opened this issue Nov 10, 2022 · 0 comments

LukeAI commented Nov 10, 2022

thanks for this demo!

I was trying to understand the implementation and I see that there is a modified softmax function
The original head uses the standard softmax implementation from pytorch:
https://github.com/Turoad/CLRNet/blob/7269e9d1c1c650343b6c7febb8e764be538b1aed/clrnet/models/heads/clr_head.py#L450

but I see that in your implementation you do something a bit different - you first subtract the max from each element in x. Why is that?

CLRNet-onnxruntime-and-tensorrt-demo/demo_trt.py

Lines 153 to 156 in 07a47e6

    
           def softmax(self, x, axis=None): 
        
               x = x - x.max(axis=axis, keepdims=True) 
        
               y = np.exp(x) 
        
               return y / y.sum(axis=axis, keepdims=True)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment