Differentiating where operator
discuss.pytorch.org
For non-NaN/inf, where(x > eps, z, y) is (x - eps > 0) * (z - y) + y which you could change to torch.sigmoid(x - eps) * T) * (z - y) + y for some tuneable float t.
1 Users
0 Comments
1 Highlights
0 Notes
Top Highlights
- For non-NaN/inf, where(x > eps, z, y) is (x - eps > 0) * (z - y) + y which you could change to torch.sigmoid(x - eps) * T) * (z - y) + y for some tuneable float t.
Domain
Ready to highlight and find good content?
Glasp is a social web highlighter that people can highlight and organize quotes and thoughts from the web, and access other like-minded people’s learning.