Merge branch 'feature/onnx_export_quant_avg_pool' of...

Merge branch 'feature/onnx_export_quant_avg_pool' of https://github.com/Xilinx/finn into feature/onnx_export_quant_avg_pool

src/finn/util/basic.py

@@ -106,6 +106,14 @@ def get_finn_root():
         )
 
 
+def get_execution_error_thresh():
+    "Return the max error that is allowed for rounding in FINN execution."
+    try:
+        return float(os.environ["ERROR_THRESH"])
+    except KeyError:
+        return 1e-2
+
+
 def make_build_dir(prefix=""):
     """Creates a temporary folder with given prefix to be used as a build dir.
     Use this function instead of tempfile.mkdtemp to ensure any generated files
@@ -305,7 +313,7 @@ def sanitize_quant_values(model, node_tensors, execution_context, check_values=F
             )
         # check if rounded values are not too far from original values
         max_error = max(np.abs(current_values - updated_values).flatten())
-        if max_error <= 1e-4:
+        if max_error <= get_execution_error_thresh():
             if check_values is True:
                 # check again if values can now be represented with set finn datatype
                 # TODO: vectorize with numpy

tests/brevitas/test_brevitas_avg_pool_export.py

@@ -55,10 +55,9 @@ def test_brevitas_avg_pool_export(
         prefix = "INT"
     else:
         prefix = "UINT"
-    dt_name = prefix + str(input_bit_width)
+    dt_name = prefix + str(input_bit_width // 2)
     dtype = DataType[dt_name]
     model = model.transform(InferShapes())
-    model.set_tensor_datatype(model.graph.input[0].name, dtype)
     model = model.transform(InferDataTypes())
 
     # execution with input tensor using integers and scale = 1
@@ -91,6 +90,7 @@ def test_brevitas_avg_pool_export(
     bo.export_finn_onnx(b_avgpool, ishape, export_onnx_path, input_t=input_quant_tensor)
     model = ModelWrapper(export_onnx_path)
     model = model.transform(InferShapes())
+    model = model.transform(InferDataTypes())
     b_avgpool.eval()
     expected = b_avgpool.forward(input_quant_tensor).tensor.detach().numpy()
     # finn execution