IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an email to s dot adaszewski at gmail dot com. User accounts are meant only to report issues and/or generate pull requests. This is a purpose-specific Git hosting for ADARED projects. Thank you for your understanding!
Vous ne pouvez pas sélectionner plus de 25 sujets Les noms de sujets doivent commencer par une lettre ou un nombre, peuvent contenir des tirets ('-') et peuvent comporter jusqu'à 35 caractères.

213 lignes
5.8KB

  1. from icosagon.decode import DEDICOMDecoder, \
  2. DistMultDecoder, \
  3. BilinearDecoder, \
  4. InnerProductDecoder
  5. import decagon_pytorch.decode.pairwise
  6. import torch
  7. def test_dedicom_decoder_01():
  8. repr_ = torch.rand(20, 32)
  9. dec_1 = DEDICOMDecoder(32, 7, keep_prob=1.,
  10. activation=torch.sigmoid)
  11. dec_2 = decagon_pytorch.decode.pairwise.DEDICOMDecoder(32, 7, drop_prob=0.,
  12. activation=torch.sigmoid)
  13. dec_2.global_interaction = dec_1.global_interaction
  14. dec_2.local_variation = dec_1.local_variation
  15. res_1 = [ dec_1(repr_, repr_, k) for k in range(7) ]
  16. res_2 = dec_2(repr_, repr_)
  17. assert isinstance(res_1, list)
  18. assert isinstance(res_2, list)
  19. assert len(res_1) == len(res_2)
  20. for i in range(len(res_1)):
  21. assert torch.all(res_1[i] == res_2[i])
  22. def test_dist_mult_decoder_01():
  23. repr_ = torch.rand(20, 32)
  24. dec_1 = DistMultDecoder(32, 7, keep_prob=1.,
  25. activation=torch.sigmoid)
  26. dec_2 = decagon_pytorch.decode.pairwise.DistMultDecoder(32, 7, drop_prob=0.,
  27. activation=torch.sigmoid)
  28. dec_2.relation = dec_1.relation
  29. res_1 = [ dec_1(repr_, repr_, k) for k in range(7) ]
  30. res_2 = dec_2(repr_, repr_)
  31. assert isinstance(res_1, list)
  32. assert isinstance(res_2, list)
  33. assert len(res_1) == len(res_2)
  34. for i in range(len(res_1)):
  35. assert torch.all(res_1[i] == res_2[i])
  36. def test_bilinear_decoder_01():
  37. repr_ = torch.rand(20, 32)
  38. dec_1 = BilinearDecoder(32, 7, keep_prob=1.,
  39. activation=torch.sigmoid)
  40. dec_2 = decagon_pytorch.decode.pairwise.BilinearDecoder(32, 7, drop_prob=0.,
  41. activation=torch.sigmoid)
  42. dec_2.relation = dec_1.relation
  43. res_1 = [ dec_1(repr_, repr_, k) for k in range(7) ]
  44. res_2 = dec_2(repr_, repr_)
  45. assert isinstance(res_1, list)
  46. assert isinstance(res_2, list)
  47. assert len(res_1) == len(res_2)
  48. for i in range(len(res_1)):
  49. assert torch.all(res_1[i] == res_2[i])
  50. def test_inner_product_decoder_01():
  51. repr_ = torch.rand(20, 32)
  52. dec_1 = InnerProductDecoder(32, 7, keep_prob=1.,
  53. activation=torch.sigmoid)
  54. dec_2 = decagon_pytorch.decode.pairwise.InnerProductDecoder(32, 7, drop_prob=0.,
  55. activation=torch.sigmoid)
  56. res_1 = [ dec_1(repr_, repr_, k) for k in range(7) ]
  57. res_2 = dec_2(repr_, repr_)
  58. assert isinstance(res_1, list)
  59. assert isinstance(res_2, list)
  60. assert len(res_1) == len(res_2)
  61. for i in range(len(res_1)):
  62. assert torch.all(res_1[i] == res_2[i])
  63. def test_is_dedicom_not_symmetric_01():
  64. repr_1 = torch.rand(20, 32)
  65. repr_2 = torch.rand(20, 32)
  66. dec = DEDICOMDecoder(32, 7, keep_prob=1.,
  67. activation=torch.sigmoid)
  68. res_1 = [ dec(repr_1, repr_2, k) for k in range(7) ]
  69. res_2 = [ dec(repr_2, repr_1, k) for k in range(7) ]
  70. assert isinstance(res_1, list)
  71. assert isinstance(res_2, list)
  72. assert len(res_1) == len(res_2)
  73. for i in range(len(res_1)):
  74. assert not torch.all(res_1[i] - res_2[i] < 0.000001)
  75. def test_is_dist_mult_symmetric_01():
  76. repr_1 = torch.rand(20, 32)
  77. repr_2 = torch.rand(20, 32)
  78. dec = DistMultDecoder(32, 7, keep_prob=1.,
  79. activation=torch.sigmoid)
  80. res_1 = [ dec(repr_1, repr_2, k) for k in range(7) ]
  81. res_2 = [ dec(repr_2, repr_1, k) for k in range(7) ]
  82. assert isinstance(res_1, list)
  83. assert isinstance(res_2, list)
  84. assert len(res_1) == len(res_2)
  85. for i in range(len(res_1)):
  86. assert torch.all(res_1[i] - res_2[i] < 0.000001)
  87. def test_is_bilinear_not_symmetric_01():
  88. repr_1 = torch.rand(20, 32)
  89. repr_2 = torch.rand(20, 32)
  90. dec = BilinearDecoder(32, 7, keep_prob=1.,
  91. activation=torch.sigmoid)
  92. res_1 = [ dec(repr_1, repr_2, k) for k in range(7) ]
  93. res_2 = [ dec(repr_2, repr_1, k) for k in range(7) ]
  94. assert isinstance(res_1, list)
  95. assert isinstance(res_2, list)
  96. assert len(res_1) == len(res_2)
  97. for i in range(len(res_1)):
  98. assert not torch.all(res_1[i] - res_2[i] < 0.000001)
  99. def test_is_inner_product_symmetric_01():
  100. repr_1 = torch.rand(20, 32)
  101. repr_2 = torch.rand(20, 32)
  102. dec = InnerProductDecoder(32, 7, keep_prob=1.,
  103. activation=torch.sigmoid)
  104. res_1 = [ dec(repr_1, repr_2, k) for k in range(7) ]
  105. res_2 = [ dec(repr_2, repr_1, k) for k in range(7) ]
  106. assert isinstance(res_1, list)
  107. assert isinstance(res_2, list)
  108. assert len(res_1) == len(res_2)
  109. for i in range(len(res_1)):
  110. assert torch.all(res_1[i] - res_2[i] < 0.000001)
  111. def test_empty_dedicom_decoder_01():
  112. repr_ = torch.rand(0, 32)
  113. dec = DEDICOMDecoder(32, 7, keep_prob=1.,
  114. activation=torch.sigmoid)
  115. res = [ dec(repr_, repr_, k) for k in range(7) ]
  116. assert isinstance(res, list)
  117. for i in range(len(res)):
  118. assert res[i].shape == (0,)
  119. def test_empty_dist_mult_decoder_01():
  120. repr_ = torch.rand(0, 32)
  121. dec = DistMultDecoder(32, 7, keep_prob=1.,
  122. activation=torch.sigmoid)
  123. res = [ dec(repr_, repr_, k) for k in range(7) ]
  124. assert isinstance(res, list)
  125. for i in range(len(res)):
  126. assert res[i].shape == (0,)
  127. def test_empty_bilinear_decoder_01():
  128. repr_ = torch.rand(0, 32)
  129. dec = BilinearDecoder(32, 7, keep_prob=1.,
  130. activation=torch.sigmoid)
  131. res = [ dec(repr_, repr_, k) for k in range(7) ]
  132. assert isinstance(res, list)
  133. for i in range(len(res)):
  134. assert res[i].shape == (0,)
  135. def test_empty_inner_product_decoder_01():
  136. repr_ = torch.rand(0, 32)
  137. dec = InnerProductDecoder(32, 7, keep_prob=1.,
  138. activation=torch.sigmoid)
  139. res = [ dec(repr_, repr_, k) for k in range(7) ]
  140. assert isinstance(res, list)
  141. for i in range(len(res)):
  142. assert res[i].shape == (0,)