-
-
Notifications
You must be signed in to change notification settings - Fork 430
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.
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
Adding the concept of training/eval mode to the VM #491
Conversation
I'm submitting this to start the discussion and understand if this feasible Without this I think the only option to disable the Dropout op in eval mode, for example, is recreating the graph which is hard when you want to create a higher level framework and still let people use gorgonia directly
Codecov Report
@@ Coverage Diff @@
## master #491 +/- ##
==========================================
+ Coverage 55.57% 55.66% +0.09%
==========================================
Files 73 74 +1
Lines 11966 12076 +110
==========================================
+ Hits 6650 6722 +72
- Misses 4443 4476 +33
- Partials 873 878 +5
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
I'm back. Still sick and sinuses still heavily infected but I think this is somewhat a good idea though I think the interface should be one method:
|
Hi! Good idea indeed. |
I was thinking about dropout since it is not a full operation so the decision can't be delayed until runtime but you're totally right eval mode is a runtime thing so it won't work anyway. I guess we should make dropout an Op |
Since the graph is only a runtime concept it doesn't make sense to have it in the graph itself.
… mode This one replaces the previous interface and unifies the methods to change the Op mode.
I made this change. please take a look again |
This implementation should be faster than previous one and also support the new TrainModeOp interface which disables it when it is not training.
I just implemented Dropout as an Op, it should be faster than previous implementation |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM...
I'm submitting this to start the discussion and understand if this feasible
Without this I think the only option to disable the Dropout op in eval mode,
for example, is recreating the graph which is hard when you want to create a
higher level framework and still let people use gorgonia directly