[Relay][OP] Fix bias_add default axis (apache#2829)

* Fix bias add default axis * update * Fix canonicalize ops for bias_add
wweic · Apr 10, 2019 · 5ce9d59 · 5ce9d59
1 parent 1ea177f
commit 5ce9d59
Show file tree

Hide file tree

Showing 5 changed files with 12 additions and 8 deletions.
diff --git a/python/tvm/relay/frontend/mxnet.py b/python/tvm/relay/frontend/mxnet.py
@@ -34,7 +34,7 @@ def _mx_fully_connected(inputs, attrs):
     res = _op.nn.dense(inputs[0], inputs[1], units=units)
     if use_bias:
         assert len(inputs) == 3
-        res = _op.nn.bias_add(res, inputs[2])
+        res = _op.nn.bias_add(res, inputs[2], axis=-1)
     return res
 
 
@@ -413,7 +413,7 @@ def _mx_batch_dot(inputs, attrs):
         raise tvm.error.OpAttributeInvalid(msg.format(transpose_a))
     if transpose_b is False:
         b = _op.transpose(b, axes=[0, 2, 1])
-    return _op.batch_matmul(a, b)
+    return _op.nn.batch_matmul(a, b)
 
 
 def _mx_arange(inputs, attrs):

diff --git a/python/tvm/relay/testing/inception_v3.py b/python/tvm/relay/testing/inception_v3.py
@@ -248,7 +248,7 @@ def get_net(batch_size,
 
     flatten = relay.nn.batch_flatten(pool)
     fc1 = relay.nn.dense(flatten, relay.var("fc1_weight"), units=num_classes)
-    fc1 = relay.nn.bias_add(fc1, relay.var("fc2_bias"))
+    fc1 = relay.nn.bias_add(fc1, relay.var("fc2_bias"), axis=-1)
     inception_v3 = relay.nn.softmax(data=fc1)
     args = relay.ir_pass.free_vars(inception_v3)
     return relay.Function(args, inception_v3)

diff --git a/python/tvm/relay/testing/layers.py b/python/tvm/relay/testing/layers.py
@@ -134,5 +134,5 @@ def dense_add_bias(data, weight=None, bias=None, units=None, **kwargs):
     if not bias:
         bias = relay.var(name + "_bias")
     data = relay.nn.dense(data, weight, units, **kwargs)
-    data = relay.nn.bias_add(data, bias)
+    data = relay.nn.bias_add(data, bias, axis=-1)
     return data
diff --git a/python/tvm/relay/testing/mlp.py b/python/tvm/relay/testing/mlp.py
@@ -50,13 +50,13 @@ def get_net(batch_size,
                      dtype=dtype)
     data = relay.nn.batch_flatten(data)
     fc1 = relay.nn.dense(data, relay.var("fc1_weight"), units=128)
-    fc1 = relay.nn.bias_add(fc1, relay.var("fc1_bias"))
+    fc1 = relay.nn.bias_add(fc1, relay.var("fc1_bias"), axis=-1)
     act1 = relay.nn.relu(fc1)
     fc2 = relay.nn.dense(act1, relay.var("fc2_weight"), units=64)
-    fc2 = relay.nn.bias_add(fc2, relay.var("fc2_bias"))
+    fc2 = relay.nn.bias_add(fc2, relay.var("fc2_bias"), axis=-1)
     act2 = relay.nn.relu(fc2)
     fc3 = relay.nn.dense(act2, relay.var("fc3_weight"), units=num_classes)
-    fc3 = relay.nn.bias_add(fc3, relay.var("fc3_bias"))
+    fc3 = relay.nn.bias_add(fc3, relay.var("fc3_bias"), axis=-1)
     mlp = relay.nn.softmax(data=fc3)
     args = relay.ir_pass.free_vars(mlp)
     return relay.Function(args, mlp)

diff --git a/src/relay/pass/canonicalize_ops.cc b/src/relay/pass/canonicalize_ops.cc
@@ -24,7 +24,11 @@ class BiasAddSimplifier : public ExprMutator {
 
       auto ttype = n->args[0]->type_as<TensorTypeNode>();
       size_t n_dim = ttype->shape.size();
-      Expr expanded_bias = ExpandBiasToMatchAxis(call->args[1], n_dim, {param->axis});
+      int axis = param->axis;
+      if (axis < 0) {
+        axis += n_dim;
+      }
+      Expr expanded_bias = ExpandBiasToMatchAxis(call->args[1], n_dim, {axis});
       Expr ret = Add(call->args[0], expanded_bias);
       ret->checked_type_ = n->checked_type_;
       return ret;