|
- # Copyright 2020 Huawei Technologies Co., Ltd
- #
- # Licensed under the Apache License, Version 2.0 (the "License");
- # you may not use this file except in compliance with the License.
- # You may obtain a copy of the License at
- #
- # http://www.apache.org/licenses/LICENSE-2.0
- #
- # Unless required by applicable law or agreed to in writing, software
- # distributed under the License is distributed on an "AS IS" BASIS,
- # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- # See the License for the specific language governing permissions and
- # limitations under the License.
-
- """inv_grad_run"""
- import numpy as np
- from tests.common.tensorio import compare_tensor
- from akg.utils import kernel_exec as utils
- from tests.common.test_op import inv_grad
- from tests.common.base import get_rtol_atol
-
-
- def inv_grad_run(shape, dtype, attrs):
- """inv_grad_run implementation"""
- if 'tuning' in attrs.keys():
- t = attrs.get("tuning", False)
- kernel_name = attrs.get("kernel_name", False)
- mod = utils.op_build_test(inv_grad.inv_grad, [shape, shape], [dtype, dtype], kernel_name=kernel_name,
- op_attrs=[], attrs=attrs, tuning=t)
- if t:
- args, exp_output, input_y, inputs_dy = gen_data(dtype, shape)
- return mod, exp_output, args
- else:
- return mod
- else:
- mod = utils.op_build_test(inv_grad.inv_grad, [shape, shape], [dtype, dtype], kernel_name='inv_grad',
- op_attrs=[], attrs=attrs)
- args, exp_output, input_y, input_dy = gen_data(dtype, shape)
- acu_output = utils.mod_launch(mod, args, expect=exp_output)
- # compare result
- rtol, atol = get_rtol_atol("inv_grad", dtype)
- testcase_result = compare_tensor(acu_output, exp_output, rtol=rtol, atol=atol, equal_nan=True)
-
- return [input_y, input_dy], acu_output, exp_output, testcase_result
-
-
- def gen_data(dtype, shape):
- # generate data
- if dtype == 'int8':
- low_bound = -128
- high_bound = 127
- elif dtype == 'int32':
- low_bound = -1000
- high_bound = 1000
- else:
- low_bound = -1.0
- high_bound = 1.0
-
- input_y = np.random.uniform(low=low_bound, high=high_bound, size=tuple(shape)).astype(dtype)
- input_dy = np.random.uniform(low=low_bound, high=high_bound, size=tuple(shape)).astype(dtype)
- if dtype in ('float16', 'int8'):
- input_y = input_y.astype("float32")
- input_dy = input_dy.astype("float32")
- exp_output = -1 * np.multiply(np.multiply(input_y, input_y), input_dy)
- if dtype in ('float16'):
- exp_output = exp_output.astype(dtype)
- input_y = input_y.astype(dtype)
- input_dy = input_dy.astype(dtype)
- elif dtype in ('int8'):
- exp_output[exp_output > 127] = 127.0
- exp_output[exp_output < -128] = -128.0
- exp_output = exp_output.astype(dtype)
- input_y = input_y.astype(dtype)
- input_dy = input_dy.astype(dtype)
-
- # inputs and output to hold the data
- output = np.full(shape, np.nan, dtype)
- args = [input_y, input_dy, output]
- return args, exp_output, input_y, input_dy
|
