[DOC, TVM] ResNet tutorial, updated TVM (#51)

vta/examples/resnet18/pynq/imagenet_predict.py

@@ -20,7 +20,7 @@ verbose = False
 # only run fpga component, mark non-conv ops as nop
 debug_fpga_only = False
 
-# Obtain model and hardware files (they're too large to check-in)
+# Obtain model files (they're too large to check-in)
 # Download them into _data dir
 data_dir = "_data/"
 url = "https://homes.cs.washington.edu/~moreau/media/vta/"
@@ -115,11 +115,6 @@ sym = vta.graph.clean_conv_fuse(sym)
 if target.device_name == "vta":
     sym = vta.graph.pack(sym, shape_dict, bfactor, cfactor)
 
-graph_attr.set_shape_inputs(sym, shape_dict)
-sym = sym.apply("InferShape")
-graph_attr.set_dtype_inputs(sym, dtype_dict)
-sym = sym.apply("InferType")
-
 with nnvm.compiler.build_config(opt_level=3):
     if target.device_name != "vta":
         graph, lib, params = nnvm.compiler.build(

vta/tests/python/integration/test_benchmark_topi_conv2d.py

@@ -38,7 +38,7 @@ def test_vta_conv2d():
         res_conv = vta.top.packed_conv2d(
             data, kernel, padding=(wl.hpad, wl.wpad), strides=(wl.hstride, wl.wstride))
         res = topi.right_shift(res_conv, 8)
-        res = topi.add(res, bias)
+        res = topi.broadcast_add(res, bias)
         res = my_clip(res, 0, 127)
         res = topi.cast(res, "int8")