Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / mathiaszinnen/focal_loss_torch issues and pull requests

#19 - weights parameter

Issue - State: open - Opened by Hossein-1991 over 1 year ago - 3 comments

#18 - Improvement for reduction="mean"

Issue - State: open - Opened by AhaTree over 1 year ago

#17 - solved an issue with a negative ignore_index (like the typically used -100 value)

Pull Request - State: closed - Opened by YoelShoshan over 1 year ago - 1 comment

#16 - problem dealing with negative ignore_index values

Issue - State: closed - Opened by YoelShoshan over 1 year ago - 2 comments

#15 - can't get the same result using other libraries

Issue - State: closed - Opened by Hossein-1991 almost 2 years ago - 4 comments

#14 - weight issue fixed, fixes #13

Pull Request - State: closed - Opened by msalhab96 almost 2 years ago

#13 - Weight computed twice

Issue - State: closed - Opened by mariusud almost 2 years ago - 3 comments

#11 - adding assertion input values, and setup.cfg for flake8

Pull Request - State: closed - Opened by msalhab96 about 2 years ago

#10 - Loss is always 0 for multiclass problem

Issue - State: closed - Opened by AngledLuffa about 2 years ago - 7 comments

#9 - Supporting Multi-class, ignore index, and different input shapes

Pull Request - State: closed - Opened by msalhab96 about 2 years ago - 1 comment

#8 - Inconsistent results

Issue - State: closed - Opened by msalhab96 about 2 years ago - 1 comment

#7 - Plans to add multiclass?

Issue - State: closed - Opened by AngledLuffa about 2 years ago - 2 comments

#6 - ignore index

Issue - State: closed - Opened by SRL94 over 2 years ago

#5 - Start a kanban board

Issue - State: closed - Opened by mathiaszinnen over 2 years ago - 2 comments

#4 - AttributeError: 'finfo' object has no attribute 'epsilon'

Issue - State: closed - Opened by talhaanwarch about 3 years ago - 1 comment

#3 - Numerical issues with torch.log(p_t)

Issue - State: closed - Opened by kaipackhaeuser about 3 years ago - 1 comment

#2 - Shouldn't there be a `(1-alpha)` pre-factor for the negatives?

Issue - State: closed - Opened by alexander-soare over 3 years ago - 2 comments

#1 - Focal Loss for Multi-class classification

Issue - State: closed - Opened by DevashishPrasad over 3 years ago - 2 comments