Some models run errors
See original GitHub issueWell , I am sorry that the question did not start a new issue and made you feel duplicated twice. This is the reason why I did not edit the question well . And that not duplicate 2 times. I’ll show you the errors separately.
As the first : when I train for dcan and np_dist . I just modify as follow(it can run well while run others model):
https://github.com/vqdang/hover_net/blob/5d1560315a3de8e7d4c8122b97b1fe9b9513910b/src/config.py#L20
https://github.com/vqdang/hover_net/blob/5d1560315a3de8e7d4c8122b97b1fe9b9513910b/src/config.py#L21
I modify it to mode=other
and self.model_type=dcan
(or mode=hover
and self.model_type=np_dist
, the two models , DCAN and NP_DIST) , and then it get the same error , such as:
it get error after finish first epoch .
Issue Analytics
- State:
- Created 3 years ago
- Comments:5 (2 by maintainers)
Top GitHub Comments
Again, there may be some unpredictable error in my environment.Or maybe I modified some of the other code, but other than those two models everything worked fine.Fcn8 and Segnet still have errors so let’s not test them.All in all, it’s important that your model is good. Thank you again for your code and paper, and for coming to your guidance these days.
I need additional info then. As I said, I tested run again by re-cloning the current repos and here are the changes The replacement of path leading to input patches is not included here. But these are enough to bootstrap the training for
fcn8
andsegnet
. I suggest you make a clean copy and double check that first.