-
Notifications
You must be signed in to change notification settings - Fork 7.1k
[video][r25d model update] fixing #1265 #1432
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
Conversation
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.
Changes are good and simplifies the code, which is great!
We should just improve on the backwards compatibility story, and then it will be good.
torchvision/models/video/resnet.py
Outdated
@@ -333,6 +340,12 @@ def r2plus1d_18(pretrained=False, progress=True, **kwargs): | |||
Returns: | |||
nn.Module: R(2+1)D-18 network | |||
""" | |||
warnings.warn( |
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.
The warning should appear when trying to load a state-dict, so you might need to have a custom _load_from_state_dict
that checks the version and maybe builds a different model depending on the version? Something similar to #1224
Codecov Report
@@ Coverage Diff @@
## master #1432 +/- ##
=========================================
Coverage ? 63.77%
=========================================
Files ? 83
Lines ? 6525
Branches ? 1009
=========================================
Hits ? 4161
Misses ? 2069
Partials ? 295
Continue to review full report at Codecov.
|
BC should be sorted - now it's only left to train the models :) |
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.
Changes LGTM, let's just wait until we have the new pre-trained weights before merging this
Hi @fmassa - did we ever got the updated models et al? |
Hi, are there any updates? This model is very important for many downstream tasks. |
I don't have the access to the models anymore (changed jobs). |
Initial PR for fixing issue 1265 regarding
a) midplanes computation on the per-layer level rather than per-block level
b) adding additional batchnorm params according to match @daniel-j-h suggestion (the impact of that on the performance is negligible, but allows easier weight transfer).
Leftover todo's: