Gluoncv SSD coco models compilation takes 4 min 30 sec on i7

I tried to run deploy_ssd_gluoncv.py tutorial code for ssd_512_resnet50_v1_coco and ssd_512_mobilenet1.0_coco models.

The compilation takes 4 min 30 sec. `target=‘llvm’, ctx=‘cpu(0)’
The last log lines before it hangs for 4 min are:

[12:34:08] /home/ubuntu/tvm/src/relay/op/type_relations.cc:111: In1:TensorType([1, 6132, 1], float32),In2:TensorType([1, 6132, 1], float32),Out:IncompleteTypeNode(0, 0x1e982ad0)

[12:34:08] /home/ubuntu/tvm/src/relay/op/type_relations.cc:111: In1:TensorType([1, 6132, 1], float32),In2:TensorType([1, 6132, 1], float32),Out:IncompleteTypeNode(0, 0x2388a0d0)

[12:34:08] /home/ubuntu/tvm/src/schedule/bound.cc:129: not in feed graph consumer = hybrid(hybrid_nms, 0x1f6ce710)
[12:34:08] /home/ubuntu/tvm/src/schedule/bound.cc:129: not in feed graph consumer = hybrid(hybrid_nms, 0x1f6ce710)