Fix (tests/quant_tensor): avoid 0 valued tensor

Xilinx · Apr 19, 2024 · 253ac7b · 253ac7b
1 parent cd121c3
commit 253ac7b
Showing 1 changed file with 2 additions and 1 deletion.
diff --git a/tests/brevitas/quant_tensor/test_quant_tensor.py b/tests/brevitas/quant_tensor/test_quant_tensor.py
@@ -39,7 +39,8 @@ def test_quant_tensor_init():
 @pytest.mark.parametrize(
     'op', [Operator.ADD, Operator.SUBTRACT, Operator.DIVIDE, Operator.MULTIPLY, Operator.MATMUL])
 def test_quant_tensor_operators(op):
-    x = torch.randn(4, 4)
+    # Avoid 0 values
+    x = 1 + torch.rand(4, 4)
 
     a = torch.Tensor(x)
     b = torch.Tensor(x)