!2682 [quant]The top level add op prefix_name check error r0.5

Merge pull request !2682 from vlne-v1/I1LJMR-quant-the-top-level-add-op-prefix_name-check-error-r0.5

mindspore/train/quant/quant.py

@@ -153,7 +153,9 @@ class ConvertToQuantNetwork:
                                                   per_channel=self.act_channel,
                                                   symmetric=self.act_symmetric,
                                                   narrow_range=self.act_range)
-            prefix = '.'.join([network.param_prefix, self._convert_op_name(prim_op.name)])
+            prefix = self._convert_op_name(prim_op.name)
+            if network.param_prefix:
+                prefix = '.'.join([network.param_prefix, self._convert_op_name(prim_op.name)])
             add_quant.update_parameters_name(prefix + '.')
             del network.__dict__[name]
             network.insert_child_to_cell(name, add_quant)

mindspore/train/quant/quant_utils.py

@@ -125,7 +125,7 @@ def scale_zp_from_fack_quant_cell(cell, data_type):
     """
     minq = cell.minq.data.asnumpy()
     maxq = cell.maxq.data.asnumpy()
-    op = cell.fake_quant
+    op = cell.fake_quant_infer
 
     scale, zp = cal_quantization_params(
         minq, maxq, data_type,

tests/ut/python/train/quant/test_quant.py

@@ -67,7 +67,7 @@ def test_qat_lenet():
     img = Tensor(np.ones((32, 1, 32, 32)).astype(np.float32))
     net = LeNet5()
     net = qat.convert_quant_network(
-        net, quant_delay=0, bn_fold=False, freeze_bn=10000, num_bits=8)
+        net, freeze_bn=10000, num_bits=8)
     # should load the checkpoint. mock here
     for param in net.get_parameters():
         param.init_data()