You are viewing a plain text version of this content. The canonical link for it is here.
Posted to discuss-archive@tvm.apache.org by fcqfcq via TVM Discuss <no...@discuss.tvm.ai> on 2020/04/07 14:59:44 UTC
[TVM Discuss] [Questions] [pytorch/onnx]model nn.conv2, in particular
dimension 1 conflicts 4096 does not match 512;
<pre> %105 = add(%104, 1f);
%106 = reshape(%105, newshape=[8, 1, 512, 1, 1]);
%107 = multiply(%3, %106);
%108 = full(1, shape=[], dtype="float32");
%109 = power(%107, 2f);
%110 = sum(%109);
%111 = add(%110, 1e-08f);
%112 = power(%111, 0.5f);
%113 = divide(%108, %112);
%114 = multiply(%113, 1f);
%115 = reshape(%114, newshape=[8, 512, 1, 1, 1]);
%116 = multiply(%107, %115);
%117 = reshape(%116, newshape=[4096, 512, 3, 3]);
%118 = nn.conv2d(%1, %117, padding=[1, 1, 1, 1], channels=4096, kernel_size=[3, 3])<font color="#CC0000"> in particular dimension 1 conflicts 4096 does not match 512; unable to unify: `Tensor[(4096, 4096, 3, 3), float32]` and `Tensor[(4096, 512, 3, 3), float32]`; </font>;
reshape(%118, newshape=[8, 512, 4, 4])
}
</pre>
---
[Visit Topic](https://discuss.tvm.ai/t/pytorch-onnx-model-nn-conv2-in-particular-dimension-1-conflicts-4096-does-not-match-512/6255/1) to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click here](https://discuss.tvm.ai/email/unsubscribe/c94b68f5899a6aa755f7611c7aba5976dd86a9674c1c17c8f29dbbc89d5dc04c).
[TVM Discuss] [Questions] [pytorch/onnx]model nn.conv2, in particular
dimension 1 conflicts 4096 does not match 512;
Posted by Zoufangyu1987 via TVM Discuss <no...@discuss.tvm.ai>.
I have the same problem. How did you solve it?
我遇到一样的问题，您是怎么解决的？
---
[Visit Topic](https://discuss.tvm.ai/t/pytorch-onnx-model-nn-conv2-in-particular-dimension-1-conflicts-4096-does-not-match-512/6255/3) to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click here](https://discuss.tvm.ai/email/unsubscribe/92e71014bcfd3fa548ee8876827e5dc4e599987821c1686d971fb033fba83e5d).
[TVM Discuss] [Questions] [pytorch/onnx]model nn.conv2, in particular
dimension 1 conflicts 4096 does not match 512;
Posted by fcqfcq via TVM Discuss <no...@discuss.tvm.ai>.
[python.onnx ] save model
%122 = take(%121, %v635, axis=0);
%123 = expand_dims(%122, axis=0);
%124 = shape_of(%blocks.1.0.weight, dtype="int64");
%125 = take(%124, %v647, axis=0);
%126 = expand_dims(%125, axis=0);
%127 = reshape(%120, newshape=[1, 4096, 4, 4]);
%128 = reshape(%blocks.1.0.weight, newshape=[1, 512, 512, 3, 3]);
%129 = multiply(%128, %v678);
%130 = take(%93, %v364, axis=1);
%131 = multiply(%blocks.1.0.fc.weight, %v655);
%132 = transpose(%131, axes=[1, 0]);
%133 = transpose(%132, axes=[1, 0]);
%134 = nn.dense(%130, %133, units=None);
%135 = reshape(%blocks.1.0.bias.bias, newshape=[1, 512]);
%136 = add(%134, %135);
%137 = add(%136, %v668);
%138 = reshape(%137, newshape=[8, 512, 1, 1, 1]);
%139 = multiply(%129, %138);
%140 = power(%139, %v691);
%141 = sum(%140, axis=[1, 3, 4]);
%142 = add(%141, %v694);
%143 = power(%142, %v696);
%144 = divide(%v698, %143);
%145 = reshape(%144, newshape=[8, 1, 512, 1, 1]);
%146 = multiply(%139, %145);
%147 = reshape(%146, newshape=[4096, 512, 3, 3]);
%148 = nn.conv2d_transpose(%127, %147, channels=512, kernel_size=[3, 3],
strides=[2, 2], padding=[0, 0, 0, 0], groups=8) in particular dimension 1 conflicts 64 does not match 512; unable to unify: `Tensor[(4096, 64, 3, 3), float32]` and
`Tensor[(4096, 512, 3, 3), float32]`; ;
%149 = shape_of(%148, dtype="int64");
%150 = take(%149, %v728, axis=0);
%151 = expand_dims(%150, axis=0);
%152 = shape_of(%148, dtype="int64");
%153 = take(%152, %v731, axis=0);
%154 = expand_dims(%153, axis=0);
%155 = (%123, %126, %151, %154);
concatenate(%155)
---
[Visit Topic](https://discuss.tvm.ai/t/pytorch-onnx-model-nn-conv2-in-particular-dimension-1-conflicts-4096-does-not-match-512/6255/2) to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click here](https://discuss.tvm.ai/email/unsubscribe/a22315253e4bad3e0e1e61963a34f0459c8c50e6e766cb5a765e36afdbd6e3ac).