-
Notifications
You must be signed in to change notification settings - Fork 3.4k
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
Faster RCNN: Unable to compile for CPU because topk fails #6340
Comments
While trying to compile it with commit a64feed, compilation goes through. |
Also, if you have a version that is already compiled, you should be able to run it with the relay VM. This is most likely an issue with the tensorflow importer. |
@ashutoshparkhi fasterrcnn should be supported, could you try with latest tvm again? very likely the topk issue you faced is from frontend, topk got one more unexpected argument |
Thanks @yongwww and @electriclilies. TOT has the fix for it. |
Hello,
This the variant that fails with topk error.
http://download.tensorflow.org/models/object_detection/faster_rcnn_inception_v2_coco_2018_01_28.tar.gz
Error:
%10934 = topk(%10930, %10933, k=None, dtype="int32")^[[31m the function is provided too many arguments expected 1, found 2; ^[[39m;
Is anyone already working to fix this issue?
Please let me know if I need to serve more details.
Tagging @yongwww as he has recently done some fixes for SSD.
Is there any plan to support this variant of Faster RCNN?
Can I try any variant that already compiles and runs with Relay VM?
Thanks,
Ashutosh
The text was updated successfully, but these errors were encountered: