jiangjiajun commented on a change in pull request #9126:
URL: https://github.com/apache/tvm/pull/9126#discussion_r717323544
##########
File path: python/tvm/relay/frontend/paddlepaddle.py
##########
@@ -138,12 +273,22 @@ def convert_conv2d(g, op, block):
kernel = g.get_node(op.input("Filter")[0])
input_x = g.get_node(op.input("Input")[0])
out_channels, _, k_h, k_w = infer_shape(kernel)
- in_h, in_w = infer_shape(input_x)[2:]
if padding_algorithm == "VALID":
paddings = [0, 0]
elif padding_algorithm == "SAME":
- pad_h = _get_pad_size(in_h, (k_h - 1) * dilations[0] + 1, strides[0])
- pad_w = _get_pad_size(in_w, (k_w - 1) * dilations[1] + 1, strides[1])
+ if strides[0] == 1 and strides[1] == 1:
+ pad_h = _get_pad_size(0, (k_h - 1) * dilations[0] + 1, strides[0])
+ pad_w = _get_pad_size(0, (k_w - 1) * dilations[1] + 1, strides[1])
+ else:
+ input_shape = shape_of(input_x)
+ h_w = _op.strided_slice(input_shape, [2], [4])
+ try:
+ in_h, in_w = infer_value(h_w, g.get_params()).numpy().tolist()
+ except Exception as e:
+ msg = "Dynamic shape is not supported in SAME padding
algorithm while stride!=1"
+ raise tvm.error.OpAttributeInvalid(msg) from e
Review comment:
Thanks, this will solve a big problem! But to avoid making this pull
request more complicated to review, let's left this for the next pull request.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscr...@tvm.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
