You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

test_batchnorm.cpp 2.6 kB

6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
6 years ago
12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182
  1. // Tencent is pleased to support the open source community by making ncnn available.
  2. //
  3. // Copyright (C) 2020 THL A29 Limited, a Tencent company. All rights reserved.
  4. //
  5. // Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
  6. // in compliance with the License. You may obtain a copy of the License at
  7. //
  8. // https://opensource.org/licenses/BSD-3-Clause
  9. //
  10. // Unless required by applicable law or agreed to in writing, software distributed
  11. // under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
  12. // CONDITIONS OF ANY KIND, either express or implied. See the License for the
  13. // specific language governing permissions and limitations under the License.
  14. #include "layer/batchnorm.h"
  15. #include "testutil.h"
  16. static int test_batchnorm(const ncnn::Mat& a, int channels, float eps)
  17. {
  18. ncnn::ParamDict pd;
  19. pd.set(0, channels); // channels
  20. pd.set(1, eps); // eps
  21. std::vector<ncnn::Mat> weights(4);
  22. weights[0] = RandomMat(channels);
  23. weights[1] = RandomMat(channels);
  24. weights[2] = RandomMat(channels);
  25. weights[3] = RandomMat(channels);
  26. // var must be positive
  27. Randomize(weights[2], 0.001f, 2.f);
  28. ncnn::Option opt;
  29. opt.num_threads = 1;
  30. opt.use_vulkan_compute = true;
  31. opt.use_int8_inference = false;
  32. int ret = test_layer<ncnn::BatchNorm>("BatchNorm", pd, weights, opt, a);
  33. if (ret != 0)
  34. {
  35. fprintf(stderr, "test_batchnorm failed a.dims=%d a=(%d %d %d) channels=%d eps=%f\n", a.dims, a.w, a.h, a.c, channels, eps);
  36. }
  37. return ret;
  38. }
  39. static int test_batchnorm_0()
  40. {
  41. return 0
  42. || test_batchnorm(RandomMat(6, 7, 16), 16, 0.f)
  43. || test_batchnorm(RandomMat(6, 7, 16), 16, 0.01f)
  44. || test_batchnorm(RandomMat(3, 5, 13), 13, 0.f)
  45. || test_batchnorm(RandomMat(3, 5, 13), 13, 0.001f);
  46. }
  47. static int test_batchnorm_1()
  48. {
  49. return 0
  50. || test_batchnorm(RandomMat(6, 16), 16, 0.f)
  51. || test_batchnorm(RandomMat(6, 16), 16, 0.01f)
  52. || test_batchnorm(RandomMat(7, 15), 15, 0.f)
  53. || test_batchnorm(RandomMat(7, 15), 15, 0.001f);
  54. }
  55. static int test_batchnorm_2()
  56. {
  57. return 0
  58. || test_batchnorm(RandomMat(128), 128, 0.f)
  59. || test_batchnorm(RandomMat(128), 128, 0.001f)
  60. || test_batchnorm(RandomMat(127), 127, 0.f)
  61. || test_batchnorm(RandomMat(127), 127, 0.1f);
  62. }
  63. int main()
  64. {
  65. SRAND(7767517);
  66. return 0
  67. || test_batchnorm_0()
  68. || test_batchnorm_1()
  69. || test_batchnorm_2();
  70. }