import torch
import numpy as np
from torch.nn import functional as F
import torch_npu
from torch_npu.testing.testcase import TestCase, run_tests
from torch_npu.testing.common_utils import create_common_tensor
class TestDropOutBackward(TestCase):
def cpu_op_exec(self, input1):
input1.requires_grad = True
out = torch.nn.Dropout(0.5)(input1)
out.backward(torch.ones_like(out))
out_grad = input1.grad
out_grad = out_grad.detach().numpy()
out = out.detach().numpy()
return out_grad, out
def npu_op_exec(self, input1):
input1.requires_grad = True
out = torch.nn.Dropout(0.5)(input1)
out.backward(torch.ones_like(out))
out_grad = input1.grad
out_grad = out_grad.to("cpu")
out_grad = out_grad.detach().numpy()
out = out.to("cpu")
out = out.detach().numpy()
return out_grad, out
def dropout_list_exec(self, list1):
epsilon = 1e-3
for item in list1:
cpu_input1, npu_input1 = create_common_tensor(item, 0, 100)
if cpu_input1.dtype == torch.float16:
cpu_input1 = cpu_input1.to(torch.float32)
cpu_output_grad, cpu_output = self.cpu_op_exec(cpu_input1)
npu_output_grad, npu_output = self.npu_op_exec(npu_input1)
cpu_output = cpu_output.astype(npu_output.dtype)
for a, b in zip(cpu_output.flatten(), npu_output.flatten()):
if abs(a) > 0 and abs(b) > 0 and abs(a - b) > epsilon:
print(f'input = {item}, ERROR!')
break
else:
print(f'input = {item}, Successfully!')
for a, b in zip(cpu_output_grad.flatten(), npu_output_grad.flatten()):
if abs(a) > 0 and abs(b) > 0 and abs(a - b) > epsilon:
print(f'input = {item}, ERROR!')
break
else:
print(f'input = {item}, Successfully!')
def test_op_shape_format_fp16(self, device="npu"):
format_list = [-1]
shape_list = [1, (32, 3, 3)]
shape_format = [
[np.float16, i, j] for i in format_list for j in shape_list
]
self.dropout_list_exec(shape_format)
def test_op_shape_format_fp32(self, device="npu"):
format_list = [-1]
shape_list = [1, (32, 3, 3)]
shape_format = [
[np.float32, i, j] for i in format_list for j in shape_list
]
self.dropout_list_exec(shape_format)
if __name__ == "__main__":
run_tests()