Add test for the qnn_add operator (#4282)

* Add test for the qnn_add operator

The tests use fake quant approach so until the tf session tensors remain in float32.
The test data has to be passed in uint8 because of how the tflite/tvm comparison works.
Abs tolerance up to 1 is allowed for the qnn results. For now input_stats are hardcoded
assuming the tests for the other qnn ops will pass the input data in the same range.

* Separate qnn uint8 test function from the fp32 elemwise tests

Isolate qnn uint8 elemwise tests
Remove blank lines

diff --git a/tests/python/frontend/tflite/test_forward.py b/tests/python/frontend/tflite/test_forward.py
index c2c3cb5..83a0730 100644 (file)
--- a/tests/python/frontend/tflite/test_forward.py
+++ b/tests/python/frontend/tflite/test_forward.py
@@ -122,7 +122,7 @@ def run_tflite_graph(tflite_model_buf, input_data):
 
 
 def compare_tflite_with_tvm(in_data, in_name, input_tensors,
-                            output_tensors, init_global_variables=False, out_names=None):
+                            output_tensors, init_global_variables=False, out_names=None, quantized=False):
     """Generic function to generate and compare TFLite and TVM output"""
     in_data = convert_to_list(in_data)
     in_name = convert_to_list(in_name)
@@ -137,6 +137,17 @@ def compare_tflite_with_tvm(in_data, in_name, input_tensors,
         # convert to tflite model
         converter = interpreter_wrapper.TFLiteConverter.from_session(
             sess, input_tensors, output_tensors)
+
+        if quantized:
+            converter.inference_type = tf.lite.constants.QUANTIZED_UINT8
+            input_arrays = converter.get_input_arrays()
+            input_stats = {}
+            # hardcode the mean_values and std_dev_values (m,s) to be the same for all inputs
+            # s = 255/(fmax-fmin);  m = -fmin*s (the zero point)
+            for i in input_arrays:
+                input_stats[i] = (128., 1.275)
+            converter.quantized_input_stats = input_stats
+
         tflite_model_buffer = converter.convert()
         tflite_output = run_tflite_graph(tflite_model_buffer, in_data)
 
@@ -148,8 +159,13 @@ def compare_tflite_with_tvm(in_data, in_name, input_tensors,
 
             tvm_output = run_tvm_graph(tflite_model_buffer, in_data, in_node, target=device,
                                        num_output=len(out_names), out_names=out_names)
-            for i in range(len(tflite_output)):
-                tvm.testing.assert_allclose(tflite_output[i], tvm_output[i], atol=1e-5, rtol=1e-5)
+            if quantized:
+                for i in range(len(tflite_output)):
+                    # allow absolute tolerance of 1 in the quantized results
+                    tvm.testing.assert_allclose(tflite_output[i], tvm_output[i], atol=1, rtol=1e-5)
+            else:
+                for i in range(len(tflite_output)):
+                    tvm.testing.assert_allclose(tflite_output[i], tvm_output[i], atol=1e-5, rtol=1e-5)
 
 
 def with_fused_activation_function(input_tensor, fn_name):
@@ -545,34 +561,53 @@ def test_forward_concatenation():
 # Element-wise
 # ---
 
-def _test_elemwise(math_op, data, fused_activation_function=None):
+def _test_elemwise(math_op, data, fused_activation_function=None, quantized=False):
     """ One iteration of elemwise """
 
     assert len(data) == 2
 
     # Test with two tensors
     with tf.Graph().as_default():
-        in_data = [array_ops.placeholder(shape=data[0].shape, dtype=data[0].dtype, name='in_0'),
-                   array_ops.placeholder(shape=data[1].shape, dtype=data[1].dtype, name='in_1')]
-        out = math_op(in_data[0], in_data[1])
-        out = with_fused_activation_function(out, fused_activation_function)
-        compare_tflite_with_tvm(data, ['in_0:0', 'in_1:0'], in_data, [out])
+        in_data = [array_ops.placeholder(shape=data[0].shape, dtype='float32', name='in_0'),
+                   array_ops.placeholder(shape=data[1].shape, dtype='float32', name='in_1')]
+
+        if quantized:
+            # fake_quant will keep the tensors in float32 until the conversion in the session
+            inq_data = [tf.quantization.fake_quant_with_min_max_args(in_data[0], min=-100, max=100, name="inq_0"),
+                        tf.quantization.fake_quant_with_min_max_args(in_data[1], min=-100, max=100, name="inq_1")]
+            out = math_op(inq_data[0], inq_data[1])
+            out = with_fused_activation_function(out, fused_activation_function)
+            out = tf.quantization.fake_quant_with_min_max_args(out, min=-200, max=200, name="out")
+            compare_tflite_with_tvm(data, ['inq_0:0', 'inq_1:0'], inq_data, [out], quantized=True)
+        else:
+            out = math_op(in_data[0], in_data[1])
+            out = with_fused_activation_function(out, fused_activation_function)
+            compare_tflite_with_tvm(data, ['in_0:0', 'in_1:0'], in_data, [out])
 
     # Test with tensor and constant
     with tf.Graph().as_default():
-        in_data = [array_ops.placeholder(shape=data[0].shape, dtype=data[0].dtype, name='in')]
-        out = math_op(in_data[0], ops.convert_to_tensor(data[1], dtype=data[1].dtype))
-        out = with_fused_activation_function(out, fused_activation_function)
-        compare_tflite_with_tvm([data[0]], ['in:0'], in_data, [out])
-
+        in_data = [array_ops.placeholder(shape=data[0].shape, dtype='float32', name='in_0')]
+
+        if quantized:
+            inq_data = [tf.quantization.fake_quant_with_min_max_args(in_data[0], min=-100, max=100, name="inq_0")]
+            inq_const = tf.quantization.fake_quant_with_min_max_args(data[1], min=-100, max=100, name="const_tensor")
+            # the 2nd tensor is treated as constant and directly added as part of the operation
+            out = math_op(inq_data, ops.convert_to_tensor(inq_const, dtype='float32', name='inq_const'))
+            out = with_fused_activation_function(out, fused_activation_function)
+            out = tf.quantization.fake_quant_with_min_max_args(out, min=-200, max=200, name="out")
+            compare_tflite_with_tvm(data[0], ['inq_0:0'], inq_data, [out], quantized=True)
+        else:
+            out = math_op(in_data[0], ops.convert_to_tensor(data[1], dtype=data[1].dtype))
+            out = with_fused_activation_function(out, fused_activation_function)
+            compare_tflite_with_tvm(data[0], ['in_0:0'], in_data, [out])
 
 #######################################################################
 # Add
 # ---
 
-def _test_add(data, fused_activation_function=None):
+def _test_add(data, fused_activation_function=None, quantized=False):
     """ One iteration of add """
-    return _test_elemwise(math_ops.add, data, fused_activation_function)
+    return _test_elemwise(math_ops.add, data, fused_activation_function, quantized)
 
 #######################################################################
 # Subtract
@@ -627,14 +662,19 @@ def _test_greater(data):
 def _test_forward_elemwise(testop):
     """ Elewise"""
     testop([np.arange(6.0, dtype=np.float32).reshape((2, 1, 1, 3)),
-               np.arange(1.0, 7.0, dtype=np.float32).reshape((2, 1, 1, 3))])
+              np.arange(1.0, 7.0, dtype=np.float32).reshape((2, 1, 1, 3))])
     testop([np.arange(6.0, dtype=np.float32).reshape((2, 1, 3)),
                np.arange(1.0, 7.0, dtype=np.float32).reshape((2, 1, 3))])
     testop([np.arange(3.0, dtype=np.float32).reshape((1, 3)),
                np.arange(1.0, 4.0, dtype=np.float32).reshape((1, 3))])
 
+def _test_forward_elemwise_quantized(testop):
+    testop([np.array(np.random.uniform(0, 255, (3, 6)), dtype=np.uint8),
+            np.array(np.random.uniform(0, 255, (3, 6)), dtype=np.uint8)], quantized=True)
+
 def test_all_elemwise():
     _test_forward_elemwise(_test_add)
+    _test_forward_elemwise_quantized(_test_add)
     _test_forward_elemwise(partial(_test_add, fused_activation_function="RELU"))
     _test_forward_elemwise(partial(_test_add, fused_activation_function="RELU6"))
     _test_forward_elemwise(_test_sub)