-
Notifications
You must be signed in to change notification settings - Fork 432
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
Fixed Conv3DTranspose with strides for data format channels_first (fixes #1714) #1748
base: main
Are you sure you want to change the base?
Conversation
tf2onnx/onnx_opset/nn.py
Outdated
expect_d = GraphBuilder(ctx).make_slice( | ||
{"data": input_shape.output[0], "ends": [4], "starts": [3], "axes": [0]}) | ||
{"data": input_shape.output[0], "ends": [sp_index_start+3], "starts": [sp_index_start+2], "axes": [0]}) |
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.
pylint is complaining that line and 533 are too long.
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.
Fixed!
onnx#1714) While shape calculations for the input correctly distinguished between channels_first and channels_last, shape calculations for the inputs of the final Slice and Pad nodes always assumed channels_last format. Signed-off-by: fthielke <[email protected]>
373519e
to
a4327e1
Compare
tf2onnx/onnx_opset/nn.py
Outdated
output_h = GraphBuilder(ctx).make_slice( | ||
{"data": output_shape.output[0], "ends": [2], "starts": [1], "axes": [0]}) | ||
{"data": output_shape.output[0], "ends": [sp_index_start+1], "starts": [sp_index_start], "axes": [0]}) |
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.
Hi @fthielke , please fix the pyline issues.
@fthielke Let me know if you have any questions. |
Signed-off-by: Jay Zhang <[email protected]>
While shape calculations for the input correctly distinguished between channels_first and channels_last, shape calculations for the inputs of the final Slice and Pad nodes always assumed channels_last format.