!1784 fix bug in fake quant grad

Merge pull request !1784 from SanjayChan/fakequant_bug_fix

mindspore/ccsrc/kernel/gpu/quant/fake_quant_grad_gpu_kernel.cc

@@ -30,7 +30,9 @@ FakeQuantGradGpuKernel::FakeQuantGradGpuKernel()
       quant_max_(0),
       quant_size_(0),
       quant_delay_(0),
-      global_step_(0) {}
+      global_step_(0),
+      narrow_range_(false),
+      symmetric_(false) {}
 
 const std::vector<size_t> &FakeQuantGradGpuKernel::GetInputSizeList() const { return input_size_list_; }
 
@@ -59,8 +61,19 @@ bool FakeQuantGradGpuKernel::Init(const CNodePtr &kernel_node) {
     MS_LOG(EXCEPTION) << "Attr \'quant_delay_\' " << quant_delay_ << " is less then 0, require larger than 0.";
   }
 
-  quant_min_ = 0;
-  quant_max_ = (1 << num_bits_) - 1;
+  symmetric_ = GetValue<bool>(AnfAlgo::GetCNodePrimitive(kernel_node)->GetAttr("symmetric"));
+  if (symmetric_) {
+    quant_min_ = 0 - (1 << (num_bits_ - 1));
+    quant_max_ = (1 << (num_bits_ - 1)) - 1;
+  } else {
+    quant_min_ = 0;
+    quant_max_ = (1 << num_bits_) - 1;
+  }
+
+  narrow_range_ = GetValue<bool>(AnfAlgo::GetCNodePrimitive(kernel_node)->GetAttr("narrow_range"));
+  if (narrow_range_) {
+    quant_min_++;
+  }
 
   if (quant_size_ == 0) {
     quant_size_ = 1;

mindspore/ccsrc/kernel/gpu/quant/fake_quant_grad_gpu_kernel.h

@@ -54,6 +54,8 @@ class FakeQuantGradGpuKernel : public GpuKernel {
   int quant_size_;
   int quant_delay_;
   int global_step_;
+  bool narrow_range_;
+  bool symmetric_;
 };
 }  // namespace kernel
 }  // namespace mindspore

mindspore/ops/_op_impl/_custom_op/fake_quant_with_min_max_grad.py

@@ -35,6 +35,8 @@ fake_quant_grad_op_info = TBERegOp("FakeQuantWithMinMaxGrad") \
     .partial_flag(True) \
     .attr("num_bits", "optional", "int", "all") \
     .attr("quant_delay", "optional", "int", "all") \
+    .attr("symmetric", "optional", "bool", "all") \
+    .attr("narrow_range", "optional", "bool", "all") \
     .input(0, "dout", None, "required", None) \
     .input(1, "x", None, "required", None) \
     .input(2, "min", None, "required", None) \
@@ -104,8 +106,9 @@ def fake_quant_with_min_max_grad_compute(dout, x, min_val, max_val, quant_min, q
     return res
 
 
-@util.check_input_type(dict, dict, dict, dict, dict, int, int, str)
-def fake_quant_with_min_max_grad(dout, x, min_val, max_val, dx, num_bits, quant_delay,
+@util.check_input_type(dict, dict, dict, dict, dict, int, int, bool, bool, str)
+def fake_quant_with_min_max_grad(dout, x, min_val, max_val, dx,
+                                 num_bits, quant_delay, symmetric, narrow_range,
                                  kernel_name="fake_quant_with_min_max_grad"):
     """FakeQuantWithMinMaxGrad"""
     input_shape = x.get("shape")
@@ -136,8 +139,15 @@ def fake_quant_with_min_max_grad(dout, x, min_val, max_val, dx, num_bits, quant_
     input_shape = (functools_reduce(lambda x, y: x * y, input_shape[:]),)
     shape_min, _, _ = util.produce_shapes(min_shape, input_shape)
 
-    quant_min = 0
-    quant_max = 2 ** num_bits - 1
+    if symmetric:
+        quant_min = 0 - 2 ** (num_bits - 1)
+        quant_max = 2 ** (num_bits - 1) - 1
+    else:
+        quant_min = 0
+        quant_max = 2 ** num_bits - 1
+    if narrow_range:
+        quant_min = quant_min + 1
+
     dout_data = tvm.placeholder(input_shape, name="dout", dtype=x_dtype)
     input_data = tvm.placeholder(input_shape, name="x", dtype=x_dtype)
     min_data = tvm.placeholder(shape_min, name="min_data", dtype=min_dtype)

mindspore/ops/_op_impl/_custom_op/fake_quant_with_min_max_update.py

@@ -23,10 +23,10 @@ from topi.cce import util
 from mindspore.ops.op_info_register import op_info_register, TBERegOp, DataType
 
 
-fake_quant_update5d_op_info = TBERegOp("FakeQuantWithMinMaxUpdate") \
+fake_quant_update_op_info = TBERegOp("FakeQuantWithMinMaxUpdate") \
     .fusion_type("OPAQUE") \
     .async_flag(False) \
-    .binfile_name("fake_quant_with_min_max_update5d.so") \
+    .binfile_name("fake_quant_with_min_max_update.so") \
     .compute_cost(10) \
     .kernel_name("fake_quant_with_min_max_update") \
     .partial_flag(True) \
@@ -47,9 +47,9 @@ fake_quant_update5d_op_info = TBERegOp("FakeQuantWithMinMaxUpdate") \
     .get_op_info()
 
 
-@op_info_register(fake_quant_update5d_op_info)
-def _fake_quant_update5d_tbe():
-    """_FakeQuantWithMinMaxUpdate5D TBE register"""
+@op_info_register(fake_quant_update_op_info)
+def _fake_quant_update_tbe():
+    """_FakeQuantWithMinMaxUpdate TBE register"""
     return
 
 

mindspore/ops/operations/_quant_ops.py

@@ -116,15 +116,17 @@ class FakeQuantWithMinMaxGrad(PrimitiveWithInfer):
         >>> _max = Tensor(np.array([2]), mindspore.float32)
         >>> result = fake_min_max_grad(dout, input_x, _min, _max)
     """
-    support_quant_bit = [4, 8]
+    support_quant_bit = [4, 7, 8]
 
     @prim_attr_register
-    def __init__(self, num_bits=8, quant_delay=0):
+    def __init__(self, num_bits=8, quant_delay=0, symmetric=False, narrow_range=False):
         if num_bits not in self.support_quant_bit:
             raise ValueError(f"For '{self.name}' attr \'num_bits\' is not support.")
 
-        self.quant_delay = validator.check_value_type('quant_delay', quant_delay, (int,), self.name)
         self.num_bits = validator.check_integer('num_bits', num_bits, 0, Rel.GT, self.name)
+        self.quant_delay = validator.check_value_type('quant_delay', quant_delay, (int,), self.name)
+        self.symmetric = validator.check_value_type('symmetric', symmetric, (bool,), self.name)
+        self.narrow_range = validator.check_value_type('narrow_range', narrow_range, (bool,), self.name)
         self.init_prim_io_names(inputs=['dout', 'x', 'min', 'max'], outputs=['dx'])
 
     def infer_shape(self, dout_shape, x_shape, min_shape, max_shape):
@@ -172,7 +174,7 @@ class FakeQuantWithMinMaxPerChannel(PrimitiveWithInfer):
         >>> _max = Tensor(np.linspace(8, 12, 12).reshape(3, 2, 2), mindspore.float32)
         >>> result = fake_quant(input_x, _min, _max)
     """
-    support_quant_bit = [4, 8]
+    support_quant_bit = [4, 7, 8]
     channel_axis = 0
 
     @prim_attr_register
@@ -219,16 +221,18 @@ class FakeQuantWithMinMaxPerChannelGrad(PrimitiveWithInfer):
         >>> _max = Tensor(np.random.randint(-2, 8, (2, 3, 4)), mindspore.float32)
         >>> result = fqmmpc_grad(dout, input_x, _min, _max)
     """
-    support_quant_bit = [4, 8]
+    support_quant_bit = [4, 7, 8]
 
     @prim_attr_register
-    def __init__(self, num_bits=8, quant_delay=0):
+    def __init__(self, num_bits=8, quant_delay=0, symmetric=False, narrow_range=False):
         """init FakeQuantWithMinMaxPerChannel Fill"""
         if num_bits not in self.support_quant_bit:
             raise ValueError(f"For '{self.name}' attr \'num_bits\' is not support.")
 
-        self.quant_delay = validator.check_value_type('quant_delay', quant_delay, (int,), self.name)
         self.num_bits = validator.check_integer('num_bits', num_bits, 0, Rel.GT, self.name)
+        self.quant_delay = validator.check_value_type('quant_delay', quant_delay, (int,), self.name)
+        self.symmetric = validator.check_value_type('symmetric', symmetric, (bool,), self.name)
+        self.narrow_range = validator.check_value_type('narrow_range', narrow_range, (bool,), self.name)
         self.init_prim_io_names(inputs=['dout', 'x', 'min', 'max'], outputs=['dx'])
 
     def infer_shape(self, dout_shape, x_shape, min_shape, max_shape):