-
Notifications
You must be signed in to change notification settings - Fork 1.2k
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
[Bug] (suggested fix) mmpose.models.pose_estimators.topdown.TopdownPoseEstimator
is unable to be symbolically traced because of untraceable add_pred_to_datasample()
and loss()
#3012
Comments
Added reproducing code and full fix suggestion |
elisa-aleman
changed the title
[Bug] mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
[Bug] [suggested fix] mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
Apr 9, 2024
elisa-aleman
changed the title
[Bug] [suggested fix] mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
[Bug] (suggested fix) mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
Apr 9, 2024
elisa-aleman
changed the title
[Bug] (suggested fix) mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
[kind/bug] (suggested fix) mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
Apr 9, 2024
elisa-aleman
changed the title
[kind/bug] (suggested fix) mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
[Bug] (suggested fix) mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
Apr 9, 2024
elisa-aleman
changed the title
[Bug] (suggested fix) mmpose.models.pose_estimators.topdown.TopdownPoseEstimator is unable to be symbolically traced because of untraceable add_pred_to_datasample() and loss()
[Bug] (suggested fix) Apr 10, 2024
mmpose.models.pose_estimators.topdown.TopdownPoseEstimator
is unable to be symbolically traced because of untraceable add_pred_to_datasample()
and loss()
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Prerequisite
Environment
computer not available at the time
Using:
torch 2.0.0+cu118
torchvision: 0.15.0+cu118
mmengine: 0.10.3
mmrazor: 1.0.0
mmpose: 1.3.1
Reproduces the problem - code sample
Using mmrazor to quantize this model, I stumbled upon an error when the
symbolic_trace
for the fx graph was being made.Applied fixes for torch 2.0.0 incompatibility suggested in mmrazor #632 and a fix for nn.Parameters inside TopdownPoseEstimator not being traced in mmrazor #633
Reproduces the problem - error message
Additional information
for
loss()
I suggest the following patch:for
add_pred_to_datasample()
I suggest the following:This solves the issue with fx tracing, although there's still other issues I have yet to solve.
The text was updated successfully, but these errors were encountered: