feat(imperative/ops): improve infer attrs validate function

GitOrigin-RevId: 6fab3b140220709c6edf92bd5a59105c96c2320a

imperative/src/impl/interpreter_impl.cpp

@@ -176,7 +176,7 @@ TensorShape ChannelImpl::get_shape(void* handle) {
     m_buffer.enqueue(Flush{info});
     m_cv.wait(lock, [&]() {
         check_worker_exc_unsafe();
-        return bool(info->ptr);
+        return static_cast<bool>(info->ptr);
     });
     m_waitee = nullptr;
     TensorShape ret = info->ptr->layout();
@@ -212,7 +212,7 @@ DeviceTensorND ChannelImpl::get_dev_tensor(void* handle) {
     m_buffer.enqueue(Flush{info});
     m_cv.wait(lock, [&]() {
         check_worker_exc_unsafe();
-        return bool(info->ptr);
+        return static_cast<bool>(info->ptr);
     });
     m_waitee = nullptr;
     return info->ptr->dev_tensor();
@@ -232,7 +232,7 @@ void ChannelImpl::close() {
 }
 
 void ChannelImpl::config_async_level(int level) {
-    mgb_assert(level <= 2 and level >= 0, "async_level should be 0, 1 or 2");
+    mgb_assert(level <= 2 && level >= 0, "async_level should be 0, 1 or 2");
     m_async_level = level;
 }
 

imperative/src/impl/ops/backward_graph.cpp

@@ -49,7 +49,7 @@ std::tuple<SmallVector<LogicalTensorDesc>, bool> BackwardGraph::InternalGraph::i
             expr_input_descs.push_back(node2attr.at(inp));
         }
 
-        auto[expr_output_descs, expr_validated] = OpDef::infer_output_attrs_fallible(
+        auto [expr_output_descs, expr_validated] = OpDef::infer_output_attrs_fallible(
             *expr_op, expr_input_descs);
         validated = validated && expr_validated;
 

imperative/src/impl/ops/batch_norm.cpp

@@ -54,16 +54,13 @@ std::tuple<SmallVector<LogicalTensorDesc>, bool> infer_output_attrs_fallible(
     SmallVector<LogicalTensorDesc> out_shapes(nr_out);
     auto&& i0 = inputs[0];
     auto&& i1 = inputs[1];
-    size_t i = 0;
-    if (!need_stat) {
-        out_shapes[0] = out_shapes[1] = {TensorLayout({0}, i0.layout.dtype, i0.layout.format), i0.comp_node};
-        i = 2;
-    }
-    for (; i < nr_out-1; ++ i) {
+    // [running_mean, running_var,] save_mean, save_var
+    for (size_t i = 0; i < nr_out-1; ++ i) {
         out_shapes[i] = {i1.layout, i1.comp_node};
     }
+    // output tensor
     out_shapes[nr_out-1] = {i0.layout, i0.comp_node};
-    return {out_shapes, true};
+    return {out_shapes, out_shapes[nr_out-1].layout.ndim != 0};
 }
 
 OP_TRAIT_REG(BatchNorm, BatchNorm, opr::BatchNorm)

imperative/src/impl/ops/broadcast.cpp

@@ -61,7 +61,7 @@ std::tuple<SmallVector<LogicalTensorDesc>, bool> infer_output_attrs_fallible(
     TensorLayout out_layout = src.layout;
     if (tshp.layout.ndim == 0 || tshp.value.empty()) {
         out_layout.ndim = 0;
-        return {{{out_layout, src.comp_node}}, true};
+        return {{{out_layout, src.comp_node}}, false};
     }
     mgb_assert(
         tshp.layout.ndim == 1,
@@ -71,7 +71,7 @@ std::tuple<SmallVector<LogicalTensorDesc>, bool> infer_output_attrs_fallible(
     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) {
+    for (size_t i = 0; i < target_ndim; ++i) {
         out_layout.shape[i] = ptr[i];
     }
     mgb_assert(valid_broadcast(src.layout, out_layout),

imperative/src/impl/ops/cond_take.cpp

@@ -76,7 +76,7 @@ std::tuple<SmallVector<LogicalTensorDesc>, bool> infer_output_attrs_fallible(
     return {{
         {TensorLayout(inputs[0].layout.dtype), cn},
         {TensorLayout(dtype::Int32()), cn}
-    }, true};
+    }, false};
 }
 
 OP_TRAIT_REG(CondTake, CondTake, opr::CondTake)

imperative/src/impl/ops/elemwise.cpp

@@ -60,7 +60,7 @@ std::tuple<SmallVector<LogicalTensorDesc>, bool> infer_output_attrs_fallible(
             TensorLayout out_layout;
             out_layout.ndim = 0;
             out_layout.dtype = out_dt;
-            return {{{out_layout, out_cn}}, true};
+            return {{{out_layout, out_cn}}, false};
         }
     }
 

imperative/src/impl/ops/tensor_manip.cpp

@@ -59,7 +59,7 @@ std::tuple<SmallVector<LogicalTensorDesc>, bool> infer_output_attrs_fallible(
     mgb_assert(inputs.size() == 1, "GetVarShape take 1 input, got %lu", inputs.size());
     auto&& desc = inputs[0];
     if (!desc.layout.ndim) {
-        return {{{TensorLayout(dtype::Int32()), desc.comp_node}}, true};
+        return {{{TensorLayout(dtype::Int32()), desc.comp_node}}, false};
     }
     DeviceTensorND value;
     if (op_def.axis == opr::GetVarShape::Param::INVALID_AXIS){