feat(imperative/ops): add infer_output_attrs for Reshape

GitOrigin-RevId: 9150d7f84d1f0a4e50f5160213c660fdca904224

imperative/src/impl/ops/broadcast.cpp

@@ -17,7 +17,7 @@
 namespace mgb {
 namespace imperative {
 
-namespace {
+namespace broadcast {
 
 std::shared_ptr<OpDef> make_from_op_node(cg::OperatorNodeBase* node_) {
     node_->cast_final_safe<opr::Broadcast>();
@@ -39,7 +39,7 @@ bool valid_broadcast(const TensorShape& src_shape,
     if (src_ndim > tar_ndim) {
         return false;
     }
-    size_t min_ndim = src_ndim < tar_ndim ? src_ndim : tar_ndim;
+    size_t min_ndim = src_ndim;
     for (size_t i = 0; i < min_ndim; ++i) {
         if (src_shape[src_ndim - i - 1] != 1 &&
             src_shape[src_ndim - i - 1] != tar_shape[tar_ndim - i - 1]) {
@@ -87,7 +87,70 @@ OP_TRAIT_REG(Broadcast, Broadcast, opr::Broadcast)
     .apply_on_var_node(apply_on_var_node)
     .infer_output_attrs_fallible(infer_output_attrs_fallible)
     .fallback();
-} // anonymous namespace
+} // broadcast
+
+namespace reshape {
+
+auto apply_on_var_node(
+        const OpDef& def,
+        const VarNodeArray& inputs) {
+    auto&& op = static_cast<const Reshape&>(def);
+    mgb_assert(inputs.size() == 2);
+    return opr::Reshape::make(inputs[0], inputs[1], op.param());
+}
+
+std::tuple<SmallVector<LogicalTensorDesc>, bool> infer_output_attrs_fallible(
+        const OpDef& def,
+        const SmallVector<LogicalTensorDesc>& inputs) {
+    auto&& op = def.cast_final_safe<Reshape>();
+    size_t nr_inp = inputs.size();
+    mgb_assert(nr_inp == 2, "Reshape expects 2 inputs; got %lu actually", nr_inp);
+    auto&& src = inputs[0];
+    auto&& tshp = inputs[1];
+
+    TensorLayout out_layout = src.layout;
+    if (tshp.layout.ndim == 0 || tshp.value.empty()) {
+        out_layout.ndim = 0;
+        return {{{out_layout, src.comp_node}}, false};
+    }
+    mgb_assert(
+        tshp.layout.ndim == 1,
+        "target shape of Broadcast expects ndim=1; got ndim=%lu actually",
+        tshp.layout.ndim);
+
+    size_t target_ndim = tshp.layout.shape[0];
+    out_layout.ndim = target_ndim;
+    auto* ptr = tshp.value.ptr<dt_int32>();
+    for (size_t i = 0; i < target_ndim; ++i) {
+        out_layout.shape[i] = ptr[i];
+    }
+
+    if (src.layout.ndim == 0) {
+        return {{{out_layout, src.comp_node}}, false};
+    }
+
+    if (op.axis != opr::Reshape::Param::INVALID_AXIS) {
+        mgb_assert(out_layout.shape[op.axis] == -1);
+        out_layout.shape[op.axis] = 1;
+        mgb_assert(src.layout.total_nr_elems() % out_layout.total_nr_elems() == 0,
+            "can not reshape from %s to %s",
+            src.layout.to_string().c_str(),
+            out_layout.to_string().c_str());
+        out_layout.shape[op.axis] = src.layout.total_nr_elems() / out_layout.total_nr_elems();
+    } else {
+        mgb_assert(src.layout.total_nr_elems() == out_layout.total_nr_elems(),
+            "can not reshape from %s to %s",
+            src.layout.to_string().c_str(),
+            out_layout.to_string().c_str());
+    }
+    return {{{out_layout, src.comp_node}}, true};
+}
+
+OP_TRAIT_REG(Reshape, Reshape)
+    .apply_on_var_node(apply_on_var_node)
+    .infer_output_attrs_fallible(infer_output_attrs_fallible)
+    .fallback();
+} // reshape
 
 }  // namespace imperative
 }  // namespace mgb

imperative/src/impl/ops/specializations.cpp

@@ -548,19 +548,6 @@ OP_TRAIT_REG(Remap, Remap)
     .fallback();
 }} // remap
 
-namespace { namespace reshape {
-auto apply_on_var_node(
-        const OpDef& def,
-        const VarNodeArray& inputs) {
-    auto&& op = static_cast<const Reshape&>(def);
-    mgb_assert(inputs.size() == 2);
-    return opr::Reshape::make(inputs[0], inputs[1], op.param());
-}
-OP_TRAIT_REG(Reshape, Reshape)
-    .apply_on_var_node(apply_on_var_node)
-    .fallback();
-}} // reshape
-
 namespace {
 auto get_index(
     const VarNodeArray& inputs, size_t vidx,