[PASS] SimplifyBatchNorm->SimplifyInference, remove dropout (#24)

apache · May 29, 2018 · 3116ef7 · 3116ef7
1 parent 02a60d0
commit 3116ef7
Show file tree

Hide file tree

Showing 4 changed files with 20 additions and 11 deletions.
diff --git a/nnvm/python/nnvm/compiler/build_module.py b/nnvm/python/nnvm/compiler/build_module.py
@@ -8,7 +8,7 @@
 from .. import runtime
 
 OPT_PASS_LEVEL = {
-    "SimplifyBatchNormInference": 2,
+    "SimplifyInference": 2,
     "PrecomputePrune": 2,
     "OpFusion": 1
 }
@@ -115,12 +115,9 @@ def optimize(graph, shape, dtype="float32"):
     """
     # pylint: disable=unused-argument
     cfg = BuildConfig.current
-    graph = graph_attr.set_shape_inputs(graph, shape)
-    graph = graph.apply("InferShape")
-    if graph.json_attr("shape_num_unknown_nodes"):
-        raise ValueError("InferShape fails..")
-    if cfg.opt_level >= OPT_PASS_LEVEL["SimplifyBatchNormInference"]:
-        graph = graph.apply("SimplifyBatchNormInference")
+    if cfg.opt_level >= OPT_PASS_LEVEL["SimplifyInference"]:
+        graph = graph_attr.set_shape_inputs(graph, shape)
+        graph = graph.apply(["InferShape", "SimplifyInference"])
     return graph
 
 

diff --git a/nnvm/python/nnvm/top/tensor.py b/nnvm/python/nnvm/top/tensor.py
@@ -44,6 +44,11 @@ def _compute(attrs, x, _):
 
 _fschedule_broadcast = tvm.convert(_schedule_broadcast)
 
+# copy
+reg.register_compute("copy", _compute_unary(topi.identity))
+reg.register_pattern("copy", OpPattern.ELEM_WISE)
+reg.register_schedule("copy", _fschedule_broadcast)
+
 # exp
 reg.register_compute("exp", _compute_unary(topi.exp))
 reg.register_pattern("exp", OpPattern.ELEM_WISE)

diff --git a/nnvm/src/compiler/simplify_batch_norm.cc → nnvm/src/compiler/simplify_inference.cc b/nnvm/src/compiler/simplify_batch_norm.cc → nnvm/src/compiler/simplify_inference.cc
@@ -22,6 +22,7 @@ BatchNormToInferUnpack(const nnvm::NodeAttrs& attrs,
                        nnvm::NodeEntry moving_mean,
                        nnvm::NodeEntry moving_var,
                        TShape dshape) {
+  CHECK_NE(dshape.ndim(), 0);
   CHECK(attrs.op);
   static const  Op* bn_op = Op::Get("batch_norm");
   CHECK(attrs.op == bn_op);
@@ -76,13 +77,14 @@ BatchNormToInferUnpack(const nnvm::NodeAttrs& attrs,
   return {out, undef, undef};
 }
 
-Graph SimplifyBatchNormInference(nnvm::Graph src) {
+Graph SimplifyInference(nnvm::Graph src) {
   // Get attributes from the graph
   const IndexedGraph& idx = src.indexed_graph();
   const ShapeVector& shape_vec = src.GetAttr<ShapeVector>("shape");
   auto transform = [&](uint32_t nid, const Node* n, std::vector<NodeEntry>* ret) {
     if (n->is_variable()) return false;
     static const Op* bn_op = Op::Get("batch_norm");
+    static const Op* dropout_op = Op::Get("dropout");
     if (n->op() == bn_op) {
       *ret = BatchNormToInferUnpack(
           n->attrs,
@@ -93,15 +95,19 @@ Graph SimplifyBatchNormInference(nnvm::Graph src) {
           n->inputs[4],
           shape_vec[idx.entry_id(nid, 0)]);
       return true;
+    } else if (n->op() == dropout_op) {
+      NodeEntry undef = MakeNode("__undef__", "undef", {});
+      *ret = {n->inputs[0], undef};
+      return true;
     } else {
       return false;
     }
   };
   return GraphTransform(src, transform);
 }
 
-NNVM_REGISTER_PASS(SimplifyBatchNormInference)
-.set_body(SimplifyBatchNormInference);
+NNVM_REGISTER_PASS(SimplifyInference)
+.set_body(SimplifyInference);
 
 }  // namespace compiler
 }  // namespace nnvm
diff --git a/...ython/compiler/test_simplify_batchnorm.py → ...ython/compiler/test_simplify_inference.py b/...ython/compiler/test_simplify_batchnorm.py → ...ython/compiler/test_simplify_inference.py
@@ -30,12 +30,13 @@ def check(dim, axis, nstep):
         for i in range(nstep):
             y1 = sym.batch_norm(
                 y1 + 1, gamma, beta, moving_mean, moving_var, epsilon=eps, axis=axis)
+            y1 = sym.dropout(y1)
             y2 = simple_bn(y2 + 1, gamma, beta, moving_mean, moving_var,
                            epsilon=eps, axis=axis, shape=ishape["x"])
         g = nnvm.graph.create(y1)
         g2 = nnvm.graph.create(y2)
         graph_attr.set_shape_inputs(g, ishape)
-        g1 = g.apply("InferShape").apply("SimplifyBatchNormInference")
+        g1 = g.apply("InferShape").apply("SimplifyInference")
         # Some prints for debug
         # print(g1.ir())
         # assert graph equals as expected