DRAFT FuseInstanceNorm with 3D 1NM shape

fix for FuseInstanceNorm with 3D 1NM shape.

Signed-off-by: SaeHie Park <[email protected]>

compiler/circle2circle-dredd-recipe-test/test.lst

@@ -63,6 +63,7 @@ Add(Net_InstanceNorm_005 PASS fuse_instnorm)
 Add(Net_InstanceNorm_006 PASS fuse_instnorm)
 Add(Net_InstanceNorm_007 PASS fuse_instnorm)
 Add(Net_InstanceNorm_008 PASS fuse_instnorm)
+Add(Net_InstanceNorm_009 PASS fuse_instnorm)
 Add(Net_Maximum_Minimum_000 PASS transform_min_max_to_relu6)
 Add(Net_Mul_Add_000 PASS remove_unnecessary_add)
 Add(Net_Mul_Add_001 PASS remove_unnecessary_add)

compiler/luci-pass-value-py-test/test.lst

@@ -42,6 +42,7 @@ eval(Net_InstanceNorm_001 fuse_instnorm)
 eval(Net_InstanceNorm_002 fuse_instnorm)
 eval(Net_InstanceNorm_003 fuse_instnorm)
 eval(Net_InstanceNorm_008 fuse_instnorm)
+eval(Net_InstanceNorm_009 fuse_instnorm)
 eval(Net_Mul_Add_000 remove_unnecessary_add)
 eval(Net_Mul_Add_001 remove_unnecessary_add)
 eval(Net_Mul_Add_002 remove_unnecessary_add)

compiler/luci/pass/src/FuseInstanceNormPass.cpp

@@ -22,6 +22,7 @@
 
 #include <luci/Profile/CircleNodeOrigin.h>
 #include <luci/Service/CircleNodeClone.h>
+#include <luci/Service/Nodes/CircleConst.h>
 
 #include <cassert>
 #include <set>
@@ -741,6 +742,12 @@ template <> bool InstanceNormPattern::match<InstanceNormPattern::PatternVersion:
   CHECK_OR_FALSE(rsqrt == rsqrt_should_be);
   CHECK_OR_FALSE(mean_of_ifm == mean_of_ifm_should_be);
 
+  // make clone for shared beta node that gets reshaped in reshape_gamma_beta()
+  auto beta_origin = luci::get_origin(const_as_beta);
+  const_as_beta = luci::clone(const_as_beta);
+  luci::add_origin(const_as_beta, beta_origin);
+  // NOTE no need to set different name as numbered suffix will be added at export
+
   // mul_gamma is absent
   // const_as_gamma assume to be 1.0
   auto graph = add_as_terminal->graph();
@@ -1075,6 +1082,11 @@ uint32_t PostFusion::input_channel(void)
   if (input_rank < 1)
     return 0;
 
+  if (input_rank == 3)
+  {
+    // use dim 1
+    return input->dim(1).value();
+  }
   // assume channel-last
   return input->dim(input_rank - 1).value();
 }

res/TensorFlowLiteRecipes/Net_InstanceNorm_009/test.recipe

@@ -0,0 +1,184 @@
+#
+# This was copied from Net_InstanceNorm_008
+# with last dim value > 1
+#
+
+operand {
+  name: "Hole"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 8
+  }
+}
+operand {
+  name: "InstanceNorm/beta"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 1
+  }
+  filler {
+    tag: "gaussian"
+    arg: "0.0"
+    arg: "1.0"
+  }
+}
+operand {
+  name: "InstanceNorm/instancenorm/add/y"
+  type: FLOAT32
+  shape {
+  }
+  filler {
+    tag: "explicit"
+    arg: "1e-06"
+  }
+}
+operand {
+  name: "InstanceNorm/moments/variance/reduction_indices"
+  type: INT32
+  shape {
+    dim: 1
+  }
+  filler {
+    tag: "explicit"
+    arg: "2"
+  }
+}
+operand {
+  name: "InstanceNorm/moments/mean"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 1
+  }
+}
+operand {
+  name: "InstanceNorm/moments/SquaredDifference"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 8
+  }
+}
+operand {
+  name: "InstanceNorm/moments/variance"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 1
+  }
+}
+operand {
+  name: "InstanceNorm/instancenorm/add"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 8
+  }
+}
+operand {
+  name: "InstanceNorm/instancenorm/Rsqrt"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 1
+  }
+}
+operand {
+  name: "InstanceNorm/instancenorm/mul_1"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 1
+  }
+}
+operand {
+  name: "InstanceNorm/instancenorm/mul_2"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 8
+  }
+}
+operand {
+  name: "InstanceNorm/instancenorm/sub"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 1
+  }
+}
+operand {
+  name: "InstanceNorm/instancenorm/add_1"
+  type: FLOAT32
+  shape {
+    dim: 1 dim: 4 dim: 8
+  }
+}
+operation {
+  type: "Mean"
+  input: "Hole"
+  input: "InstanceNorm/moments/variance/reduction_indices"
+  output: "InstanceNorm/moments/mean"
+  mean_options {
+    keep_dims: true
+  }
+}
+operation {
+  type: "SquaredDifference"
+  input: "Hole"
+  input: "InstanceNorm/moments/mean"
+  output: "InstanceNorm/moments/SquaredDifference"
+}
+operation {
+  type: "Mean"
+  input: "InstanceNorm/moments/SquaredDifference"
+  input: "InstanceNorm/moments/variance/reduction_indices"
+  output: "InstanceNorm/moments/variance"
+  mean_options {
+    keep_dims: true
+  }
+}
+operation {
+  type: "Add"
+  input: "InstanceNorm/moments/variance"
+  input: "InstanceNorm/instancenorm/add/y"
+  output: "InstanceNorm/instancenorm/add"
+  add_options {
+    activation: NONE
+  }
+}
+operation {
+  type: "Rsqrt"
+  input: "InstanceNorm/instancenorm/add"
+  output: "InstanceNorm/instancenorm/Rsqrt"
+}
+operation {
+  type: "Mul"
+  input: "Hole"
+  input: "InstanceNorm/instancenorm/Rsqrt"
+  output: "InstanceNorm/instancenorm/mul_1"
+  mul_options {
+    activation: NONE
+  }
+}
+operation {
+  type: "Mul"
+  input: "InstanceNorm/moments/mean"
+  input: "InstanceNorm/instancenorm/Rsqrt"
+  output: "InstanceNorm/instancenorm/mul_2"
+  mul_options {
+    activation: NONE
+  }
+}
+operation {
+  type: "Sub"
+  input: "InstanceNorm/beta"
+  input: "InstanceNorm/instancenorm/mul_2"
+  output: "InstanceNorm/instancenorm/sub"
+  sub_options {
+    activation: NONE
+  }
+}
+operation {
+  type: "Add"
+  input: "InstanceNorm/instancenorm/mul_1"
+  input: "InstanceNorm/instancenorm/sub"
+  output: "InstanceNorm/instancenorm/add_1"
+  add_options {
+    activation: NONE
+  }
+}
+input: "Hole"
+output: "InstanceNorm/instancenorm/add_1"

res/TensorFlowLiteRecipes/Net_InstanceNorm_009/test.rule

@@ -0,0 +1,13 @@
+# To check if this network is converted to circle InstanceNorm op
+
+RULE    "VERIFY_FILE_FORMAT"      $(verify_file_format) '=' 1
+
+RULE    "INSTANCE_NORM_EXIST"     $(op_count INSTANCE_NORM) '=' 1
+RULE    "NO_ADD"                  $(op_count ADD) '=' 0
+RULE    "NO_MUL"                  $(op_count MUL) '=' 0
+RULE    "NO_POW"                  $(op_count POW) '=' 0
+RULE    "NO_DIV"                  $(op_count DIV) '=' 0
+RULE    "NO_SQUARED_DIFF"         $(op_count SQUARED_DIFFERENCE) '=' 0
+RULE    "NO_MEAN"                 $(op_count MEAN) '=' 0
+RULE    "NO_RSQRT"                $(op_count RSQRT) '=' 0
+RULE    "NO_SUB"                  $(op_count SUB) '=' 0