Delete folder ViNDr after moving to output/

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/abnormal_error_top_100_sent_diff_emb.txt

@@ -1,100 +0,0 @@
-1. upon close examination , these calcifications appear to be stable in size and number but appear more prominent in the cc projection
-2. stable benign appearing calcifications are seen in both breasts , to include vascular calcifications
-3. benign - appearing calcifications continues to show a slow progressive increase in number
-4. stable benign appearing calcifications are seen in both breasts , to include vascular calcification
-5. possibly new , nonspecific linear calcifications are now seen in the posteroinferior aspect of the right breast , specifically on the synthetic 2 - d reconstruction
-6. progressive predominantly linear segmentally distributed calcifications are noted in the medial right breast
-7. there are stable adjacent benign - appearing calcifications
-8. a biopsy clip and surgical clips are present on the right along with benign - appearing calcifications
-9. few benign calcifications are also seen bilaterally , without significant change
-10. few benign calcifications are present bilaterally without significant change
-11. there are a few associated stable residual calcifications
-12. there are stable scattered benign - appearing calcifications in both breasts
-13. there are evolving post reduction changes including multiple oil cysts and calcifications
-14. there are scattered benign calcifications in both breasts and reticular radiopaque material identified within otherwise stable appearing skin lesions , presumably moles
-15. there are stable benign - appearing calcifications in both breasts
-16. benign - appearing calcified fat necrosis is a stable in the left breast
-17. there are scattered benign - appearing calcifications and calcified oil cysts bilaterally
-18. stable benign - appearing nodules and calcifications are present on the left
-19. stable benign nodules and calcifications are present bilaterally
-20. findings consistent with bilateral retroareolar calcific fat necrosis secondary to bilateral breast reduction surgery is again noted without significant interval change
-21. a few scattered calcifications again seen bilaterally when compared to the prior exam ( s ), there has been no significant interval change
-22. there is been no interval significant change in the multiple circumscribed nodules seen bilaterally as well as the coarse calcifications in both breasts
-23. there are extensive capsular calcifications on the left and mild capsular calcifications on the right , stable
-24. interval calcifications at the operative bed
-25. architectural distortion and dystrophic calcification in the left breast remain stable
-26. there is a stable benign nodule and benign - appearing calcifications
-27. these calcifications have increased since previous studies dating back to [ date ] although they have more significantly increased over the last several years
-28. there are adjacent stable benign - appearing large rodlike calcifications
-29. multiple bilateral scattered stable similar appearing nodules and multiple bilateral scattered secretory calcifications are considered benign
-30. a few isolated typically benign calcifications are seen bilaterally
-31. there are a few scattered calcifications , mildly progressive , but with a benign appearance
-32. no suspicious findings or significant changes are present as compared to the prior examination ( s ), the most recent dated [ date ] except that several small benign appearing calcifications are noted on the right new since the previous study
-33. a few benign - appearing calcifications again seen
-34. when compared to the prior exam ( s ), there has been progressive benign - appearing calcifications
-35. a large benign dystrophic calcification in the retroareolar right breast is present
-36. there are interval grouped calcifications at the operative bed
-37. a few small stable benign calcifications are present in both breasts
-38. there are a few scattered benign - appearing calcifications as well as vascular calcifications
-39. no suspicious findings or significant changes are present as compared to the prior examination ( s ), the most recent dated [ date ] except for the calcifications are more visualized on the current study
-40. the calcifications again seen
-41. there are a few calcifications in both breasts , mildly progressive but with a benign appearance
-42. there are bilateral stable scattered benign - appearing calcifications
-43. bilateral nodule stable benign appearing calcifications are seen in both breasts
-44. progressive grouped calcifications in the left breast
-45. benign secretory calcifications can be identified bilaterally
-46. there are a few bilateral stable benign - appearing calcifications
-47. there are a few scattered calcifications , mildly progressive but with a benign appearance
-48. secretory calcifications and calcified oil cysts are again scattered throughout both breasts
-49. small stable appearing left - sided calcifications are again noted
-50. multiple benign calcifications are present bilaterally without significant a biopsy clip in the left breast is again seen
-51. there are stable benign - appearing calcifications bilaterally
-52. bilateral benign - appearing stable oil cyst and calcifications are noted
-53. there are innumerable bilateral benign - appearing calcifications
-54. there are multiple scattered benign - appearing calcifications and a few scattered benign - appearing densities bilaterally
-55. there are mildly progressive bilateral calcifications which have a benign appearance
-56. several benign - appearing calcifications are noted on the right
-57. post - surgical changes are noted in the left breast and left axilla with clips and dystrophic calcifications noted
-58. tiny stable calcifications are noted on the left
-59. bilateral nodule stable benign appearing calcifications are seen in the left breast
-60. a few benign stable calcifications are present in both breasts
-61. numerous scattered benign type calcifications can be identified in all 4 quadrants bilaterally
-62. a few scattered calcifications are again seen bilaterally , mildly progressive but with a benign appearance
-63. a few scattered calcifications are again seen
-64. a few small stable benign - appearing calcifications are present in both breasts
-65. a few stable benign - appearing calcifications are noted
-66. stable nodular pattern with scattered coarse calcifications
-67. a few benign - appearing calcifications are noted bilaterally
-68. a few small stable and benign - appearing calcifications are present in both breasts
-69. there are scattered small calcifications , mildly progressive but with a benign appearance
-70. expected benign features of evolving fat necrosis are evident , including oral cyst formation and rare coarse calcification
-71. benign calcifications are present bilaterally without significant change
-72. benign retroareolar calcifications can again be identified bilaterally
-73. stable benign appearing left breast calcifications
-74. there are a few scattered calcifications , progressive , but with a benign appearance
-75. a benign stable dystrophic calcification is in the upper outer left breast at mid depth
-76. benign - appearing calcifications and nodules are present bilaterally
-77. there are a few scattered calcifications bilaterally which are mildly progressive but have a benign appearance
-78. bilateral scattered stable nodules and a few bilateral diffusely scattered punctate calcifications are again seen
-79. there are a few stable calcifications bilaterally
-80. bilateral benign / stable postsurgical change , and benign calcification
-81. a few benign - appearing calcifications are present
-82. small stable calcifications are noted bilaterally
-83. scattered benign type calcifications can be identified bilaterally
-84. there are scattered benign - appearing calcifications
-85. stable benign calcifications and nodularity is are once again noted bilaterally
-86. there are a few scattered calcifications in both breasts , mildly progressive but with a benign appearance
-87. there is a interval calcification in the operative bed on the left which has a dystrophic benign appearance
-88. a few benign - appearing calcifications are present bilaterally
-89. a calcified oil cyst is noted on the left
-90. stable small calcifications are noted bilaterally
-91. stable diffuse benign - appearing calcifications are present in the right breast
-92. there are stable benign - appearing large rodlike calcifications within both breasts
-93. a few small stable benign calcifications are present in the left breast
-94. postoperative changes and benign calcifications are stable in both breasts
-95. a few calcifications again seen bilaterally , mildly progressive but with a benign appearance
-96. there are several scattered benign - appearing calcifications
-97. left breast progressive calcifications as described above
-98. a couple of benign calcifications are present bilaterally
-99. stable benign calcifications are present on the left
-100. bilateral nodule stable benign appearing calcifications are seen in the right breast

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/abnormal_hypothesis_dict.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:883e696d219d713875e7b96632b9a5956740ac2e588f941b47385f999b5429b9
-size 435

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/abnormal_prompt_dict.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3482e6c48f995549a361be83eea7adfedd8bf7ec11c69092bec30d81dd6ab75f
-size 1350

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/aligner_30.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:172a1878c7a48f044165c0bd29a27d38f361a6233a8dd72a66668ce6bd5a5f9f
-size 4197802

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/aligner_out.txt

@@ -1,76 +0,0 @@
-2024-11-06 17:13:42,168 - Train size: classifier [(11473, 2048)], clip [(11473, 512)]
-2024-11-06 17:13:42,168 - Valid size: classifier [(4918, 2048)], clip [(4918, 512)]
-2024-11-06 17:13:42,168 - Training linear aligner ...
-2024-11-06 17:13:42,168 - Linear alignment train: ((11473, 2048)) --> ((11473, 512)).
-2024-11-06 17:13:42,168 - Linear alignment test: ((4918, 2048)) --> ((4918, 512)).
-2024-11-06 17:13:44,094 - Initial MSE, R^2: 6.072, -0.349
-2024-11-06 17:13:45,927 - Epoch number, 0, train loss: 0.901, test MSE: 0.392, test_r2: 0.913, best MSE: 0.392
-2024-11-06 17:13:46,337 - Epoch number, 1, train loss: 0.386, test MSE: 0.336, test_r2: 0.925, best MSE: 0.336
-2024-11-06 17:13:46,762 - Epoch number, 2, train loss: 0.338, test MSE: 0.302, test_r2: 0.933, best MSE: 0.302
-2024-11-06 17:13:47,163 - Epoch number, 3, train loss: 0.310, test MSE: 0.281, test_r2: 0.938, best MSE: 0.281
-2024-11-06 17:13:47,580 - Epoch number, 4, train loss: 0.292, test MSE: 0.270, test_r2: 0.940, best MSE: 0.270
-2024-11-06 17:13:48,007 - Epoch number, 5, train loss: 0.279, test MSE: 0.258, test_r2: 0.943, best MSE: 0.258
-2024-11-06 17:13:48,431 - Epoch number, 6, train loss: 0.269, test MSE: 0.256, test_r2: 0.943, best MSE: 0.256
-2024-11-06 17:13:48,864 - Epoch number, 7, train loss: 0.259, test MSE: 0.243, test_r2: 0.946, best MSE: 0.243
-2024-11-06 17:13:49,254 - Epoch number, 8, train loss: 0.252, test MSE: 0.238, test_r2: 0.947, best MSE: 0.238
-2024-11-06 17:13:49,665 - Epoch number, 9, train loss: 0.247, test MSE: 0.229, test_r2: 0.949, best MSE: 0.229
-2024-11-06 17:13:50,085 - Epoch number, 10, train loss: 0.241, test MSE: 0.228, test_r2: 0.949, best MSE: 0.228
-2024-11-06 17:13:50,512 - Epoch number, 11, train loss: 0.236, test MSE: 0.224, test_r2: 0.950, best MSE: 0.224
-2024-11-06 17:13:50,918 - Epoch number, 12, train loss: 0.232, test MSE: 0.219, test_r2: 0.951, best MSE: 0.219
-2024-11-06 17:13:51,308 - Epoch number, 13, train loss: 0.228, test MSE: 0.219, test_r2: 0.951, best MSE: 0.219
-2024-11-06 17:13:51,720 - Epoch number, 14, train loss: 0.224, test MSE: 0.214, test_r2: 0.952, best MSE: 0.214
-2024-11-06 17:13:52,141 - Epoch number, 15, train loss: 0.222, test MSE: 0.210, test_r2: 0.953, best MSE: 0.210
-2024-11-06 17:13:52,554 - Epoch number, 16, train loss: 0.218, test MSE: 0.207, test_r2: 0.954, best MSE: 0.207
-2024-11-06 17:13:52,946 - Epoch number, 17, train loss: 0.216, test MSE: 0.206, test_r2: 0.954, best MSE: 0.206
-2024-11-06 17:13:53,372 - Epoch number, 18, train loss: 0.214, test MSE: 0.209, test_r2: 0.954, best MSE: 0.206
-2024-11-06 17:13:53,789 - Epoch number, 19, train loss: 0.212, test MSE: 0.202, test_r2: 0.955, best MSE: 0.202
-2024-11-06 17:13:54,206 - Epoch number, 20, train loss: 0.210, test MSE: 0.199, test_r2: 0.956, best MSE: 0.199
-2024-11-06 17:13:54,617 - Epoch number, 21, train loss: 0.207, test MSE: 0.198, test_r2: 0.956, best MSE: 0.198
-2024-11-06 17:13:55,022 - Epoch number, 22, train loss: 0.206, test MSE: 0.195, test_r2: 0.957, best MSE: 0.195
-2024-11-06 17:13:55,435 - Epoch number, 23, train loss: 0.204, test MSE: 0.194, test_r2: 0.957, best MSE: 0.194
-2024-11-06 17:13:55,856 - Epoch number, 24, train loss: 0.202, test MSE: 0.195, test_r2: 0.957, best MSE: 0.194
-2024-11-06 17:13:56,273 - Epoch number, 25, train loss: 0.202, test MSE: 0.192, test_r2: 0.957, best MSE: 0.192
-2024-11-06 17:13:56,701 - Epoch number, 26, train loss: 0.200, test MSE: 0.191, test_r2: 0.958, best MSE: 0.191
-2024-11-06 17:13:57,107 - Epoch number, 27, train loss: 0.198, test MSE: 0.189, test_r2: 0.958, best MSE: 0.189
-2024-11-06 17:13:57,518 - Epoch number, 28, train loss: 0.197, test MSE: 0.187, test_r2: 0.958, best MSE: 0.187
-2024-11-06 17:13:57,947 - Epoch number, 29, train loss: 0.196, test MSE: 0.194, test_r2: 0.957, best MSE: 0.187
-2024-11-06 17:13:57,960 - Aligner weights saved to /restricted/projectnb/batmanlab/shawn24/PhD/Ladder/out/ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/aligner_30.pth
-2024-11-06 17:13:57,960 - Saved aligner to /restricted/projectnb/batmanlab/shawn24/PhD/Ladder/out/ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect
-2024-11-06 18:57:23,915 - Train size: classifier [(11473, 2048)], clip [(11473, 512)]
-2024-11-06 18:57:23,916 - Valid size: classifier [(4918, 2048)], clip [(4918, 512)]
-2024-11-06 18:57:23,916 - Training linear aligner ...
-2024-11-06 18:57:23,916 - Linear alignment train: ((11473, 2048)) --> ((11473, 512)).
-2024-11-06 18:57:23,916 - Linear alignment test: ((4918, 2048)) --> ((4918, 512)).
-2024-11-06 18:57:25,585 - Initial MSE, R^2: 5.838, -0.297
-2024-11-06 18:57:26,085 - Epoch number, 0, train loss: 1.001, test MSE: 0.454, test_r2: 0.899, best MSE: 0.454
-2024-11-06 18:57:26,498 - Epoch number, 1, train loss: 0.441, test MSE: 0.377, test_r2: 0.916, best MSE: 0.377
-2024-11-06 18:57:26,900 - Epoch number, 2, train loss: 0.379, test MSE: 0.338, test_r2: 0.925, best MSE: 0.338
-2024-11-06 18:57:27,300 - Epoch number, 3, train loss: 0.343, test MSE: 0.314, test_r2: 0.930, best MSE: 0.314
-2024-11-06 18:57:27,714 - Epoch number, 4, train loss: 0.319, test MSE: 0.294, test_r2: 0.935, best MSE: 0.294
-2024-11-06 18:57:28,094 - Epoch number, 5, train loss: 0.301, test MSE: 0.281, test_r2: 0.938, best MSE: 0.281
-2024-11-06 18:57:28,502 - Epoch number, 6, train loss: 0.288, test MSE: 0.270, test_r2: 0.940, best MSE: 0.270
-2024-11-06 18:57:28,913 - Epoch number, 7, train loss: 0.276, test MSE: 0.261, test_r2: 0.942, best MSE: 0.261
-2024-11-06 18:57:29,314 - Epoch number, 8, train loss: 0.267, test MSE: 0.256, test_r2: 0.943, best MSE: 0.256
-2024-11-06 18:57:29,720 - Epoch number, 9, train loss: 0.259, test MSE: 0.247, test_r2: 0.945, best MSE: 0.247
-2024-11-06 18:57:30,125 - Epoch number, 10, train loss: 0.252, test MSE: 0.243, test_r2: 0.946, best MSE: 0.243
-2024-11-06 18:57:30,540 - Epoch number, 11, train loss: 0.246, test MSE: 0.238, test_r2: 0.947, best MSE: 0.238
-2024-11-06 18:57:30,947 - Epoch number, 12, train loss: 0.241, test MSE: 0.234, test_r2: 0.948, best MSE: 0.234
-2024-11-06 18:57:31,333 - Epoch number, 13, train loss: 0.237, test MSE: 0.230, test_r2: 0.949, best MSE: 0.230
-2024-11-06 18:57:31,728 - Epoch number, 14, train loss: 0.232, test MSE: 0.227, test_r2: 0.950, best MSE: 0.227
-2024-11-06 18:57:32,137 - Epoch number, 15, train loss: 0.229, test MSE: 0.223, test_r2: 0.950, best MSE: 0.223
-2024-11-06 18:57:32,560 - Epoch number, 16, train loss: 0.225, test MSE: 0.221, test_r2: 0.951, best MSE: 0.221
-2024-11-06 18:57:32,973 - Epoch number, 17, train loss: 0.222, test MSE: 0.218, test_r2: 0.952, best MSE: 0.218
-2024-11-06 18:57:33,384 - Epoch number, 18, train loss: 0.219, test MSE: 0.216, test_r2: 0.952, best MSE: 0.216
-2024-11-06 18:57:33,793 - Epoch number, 19, train loss: 0.216, test MSE: 0.214, test_r2: 0.952, best MSE: 0.214
-2024-11-06 18:57:34,197 - Epoch number, 20, train loss: 0.213, test MSE: 0.212, test_r2: 0.953, best MSE: 0.212
-2024-11-06 18:57:34,595 - Epoch number, 21, train loss: 0.211, test MSE: 0.210, test_r2: 0.953, best MSE: 0.210
-2024-11-06 18:57:34,992 - Epoch number, 22, train loss: 0.209, test MSE: 0.208, test_r2: 0.954, best MSE: 0.208
-2024-11-06 18:57:35,406 - Epoch number, 23, train loss: 0.207, test MSE: 0.206, test_r2: 0.954, best MSE: 0.206
-2024-11-06 18:57:35,820 - Epoch number, 24, train loss: 0.205, test MSE: 0.205, test_r2: 0.955, best MSE: 0.205
-2024-11-06 18:57:36,233 - Epoch number, 25, train loss: 0.203, test MSE: 0.204, test_r2: 0.955, best MSE: 0.204
-2024-11-06 18:57:36,645 - Epoch number, 26, train loss: 0.201, test MSE: 0.202, test_r2: 0.955, best MSE: 0.202
-2024-11-06 18:57:37,060 - Epoch number, 27, train loss: 0.199, test MSE: 0.202, test_r2: 0.955, best MSE: 0.202
-2024-11-06 18:57:37,483 - Epoch number, 28, train loss: 0.198, test MSE: 0.199, test_r2: 0.956, best MSE: 0.199
-2024-11-06 18:57:37,890 - Epoch number, 29, train loss: 0.197, test MSE: 0.198, test_r2: 0.956, best MSE: 0.198
-2024-11-06 18:57:37,902 - Aligner weights saved to /restricted/projectnb/batmanlab/shawn24/PhD/Ladder/out/ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/aligner_30.pth
-2024-11-06 18:57:37,902 - Saved aligner to /restricted/projectnb/batmanlab/shawn24/PhD/Ladder/out/ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/all_slices_y_ensemble_yes_H1_benign calcifications.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:4ed9d1567eb9fe3964cba2251896f788054e6312885809b967593b2d17d1feb9
-size 10138

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/all_slices_y_ensemble_yes_H2_stable calcifications.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:6f4f813b9adaf40514fc697d83d5eba2d3adb5468a279bf66941037d62c8ffdd
-size 10138

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/all_slices_y_ensemble_yes_H3_progressive calcifications.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a6960a67625ed0e0d9bce4c6e01d3dde8418a1a3cba4a778afc53242815c96df
-size 10296

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/all_slices_y_ensemble_yes_H4_scattered calcifications.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1bea597f880caba9d917b0371a99effb1f219d1f81fabbe1e2428bf4a7c8590f
-size 10156

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/all_slices_y_ensemble_yes_H5_nodules.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:d01e4d2aff2aef1684c6d7876f59bb51b8aa871d54e8e06fede1c90dfa589e03
-size 10054

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/ladder_discover_slices_performance_ERM.txt

@@ -1,8 +0,0 @@
-Accuracy for Cancer patients with calc: 0.9038461538461539
-Accuracy for Cancer patients without  calc (Worst Group): 0.6747572815533981
-Accuracy for Cancer overall patients: 0.7516129032258064
-
-
-AUROC for overall (Mean): 0.8000759548611112
-AUROC for positive disease with  calc vs all negatives: 0.9180209168002136
-AUROC for positive disease without  calc vs all negatives: 0.7405309255326322

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/ladder_mitigate_slices.txt

@@ -1,8 +0,0 @@
-Accuracy for Cancer patients with calc: 0.9607843137254902
-Accuracy for Cancer patients without  calc (Worst Group): 0.8655913978494624
-Accuracy for Cancer overall patients: 0.8993055555555556
-
-
-AUROC for overall (Mean): 0.8848001123661093
-AUROC for positive disease with  calc vs all negatives: 0.9418949612942104
-AUROC for positive disease without  calc vs all negatives: 0.8534900339216668

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/ladder_validate_slices_w_LLM-abnormal.txt

@@ -1,30 +0,0 @@
-Hypothesis Dictionary:
-{'H1': 'The classifier is making mistake as it is biased toward benign calcifications', 'H2': 'The classifier is making mistake as it is biased toward stable calcifications', 'H3': 'The classifier is making mistake as it is biased toward progressive calcifications', 'H4': 'The classifier is making mistake as it is biased toward scattered calcifications', 'H5': 'The classifier is making mistake as it is biased toward nodules'}
-
-Prompt Dictionary:
-{'H1_benign calcifications': ['The image shows benign calcifications.', 'Identify benign calcifications in the mammogram.', 'This mammogram contains benign calcifications.', 'Detect benign calcifications in the breast image.', 'Benign calcifications are present in this scan.'], 'H2_stable calcifications': ['The image shows stable calcifications.', 'Identify stable calcifications in the mammogram.', 'This mammogram contains stable calcifications.', 'Detect stable calcifications in the breast image.', 'Stable calcifications are present in this scan.'], 'H3_progressive calcifications': ['The image shows progressive calcifications.', 'Identify progressive calcifications in the mammogram.', 'This mammogram contains progressive calcifications.', 'Detect progressive calcifications in the breast image.', 'Progressive calcifications are present in this scan.'], 'H4_scattered calcifications': ['The image shows scattered calcifications.', 'Identify scattered calcifications in the mammogram.', 'This mammogram contains scattered calcifications.', 'Detect scattered calcifications in the breast image.', 'Scattered calcifications are present in this scan.'], 'H5_nodules': ['The image shows nodules.', 'Identify nodules in the mammogram.', 'This mammogram contains nodules.', 'Detect nodules in the breast image.', 'Nodules are present in this scan.']}
-==============================================
-0 H1_benign calcifications
-Accuracy on the error slice (where attribute absent, the hypothesis failed): 0.58
-Accuracy on the bias aligned slice (where attribute present, the hypothesis passed): 0.9327731092436975
-==============================================
-==============================================
-1 H2_stable calcifications
-Accuracy on the error slice (where attribute absent, the hypothesis failed): 0.5714285714285714
-Accuracy on the bias aligned slice (where attribute present, the hypothesis passed): 0.9330543933054394
-==============================================
-==============================================
-2 H3_progressive calcifications
-Accuracy on the error slice (where attribute absent, the hypothesis failed): 0.6111111111111112
-Accuracy on the bias aligned slice (where attribute present, the hypothesis passed): 0.9316239316239316
-==============================================
-==============================================
-3 H4_scattered calcifications
-Accuracy on the error slice (where attribute absent, the hypothesis failed): 0.6
-Accuracy on the bias aligned slice (where attribute present, the hypothesis passed): 0.9285714285714286
-==============================================
-==============================================
-4 H5_nodules
-Accuracy on the error slice (where attribute absent, the hypothesis failed): 0.8863636363636364
-Accuracy on the bias aligned slice (where attribute present, the hypothesis passed): 0.8589743589743589
-==============================================

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/prompt.txt

@@ -1,141 +0,0 @@
-
-    Context: Breast cancer classification from mammograms using a deep neural network
-    Analysis post training: On a validation set, 
-    a. Get the difference between the image embeddings of correct and incorrectly classified samples to estimate the features present in the correctly classified samples but missing in the misclassified samples.
-    b. Retrive the top 50 sentences from radiology report that matches closely to the embedding difference in step a.
-    c. The sentence list is given below:
-    1. upon close examination , these calcifications appear to be stable in size and number but appear more prominent in the cc projection
-2. stable benign appearing calcifications are seen in both breasts , to include vascular calcifications
-3. benign - appearing calcifications continues to show a slow progressive increase in number
-4. stable benign appearing calcifications are seen in both breasts , to include vascular calcification
-5. possibly new , nonspecific linear calcifications are now seen in the posteroinferior aspect of the right breast , specifically on the synthetic 2 - d reconstruction
-6. progressive predominantly linear segmentally distributed calcifications are noted in the medial right breast
-7. there are stable adjacent benign - appearing calcifications
-8. a biopsy clip and surgical clips are present on the right along with benign - appearing calcifications
-9. few benign calcifications are also seen bilaterally , without significant change
-10. few benign calcifications are present bilaterally without significant change
-11. there are a few associated stable residual calcifications
-12. there are stable scattered benign - appearing calcifications in both breasts
-13. there are evolving post reduction changes including multiple oil cysts and calcifications
-14. there are scattered benign calcifications in both breasts and reticular radiopaque material identified within otherwise stable appearing skin lesions , presumably moles
-15. there are stable benign - appearing calcifications in both breasts
-16. benign - appearing calcified fat necrosis is a stable in the left breast
-17. there are scattered benign - appearing calcifications and calcified oil cysts bilaterally
-18. stable benign - appearing nodules and calcifications are present on the left
-19. stable benign nodules and calcifications are present bilaterally
-20. findings consistent with bilateral retroareolar calcific fat necrosis secondary to bilateral breast reduction surgery is again noted without significant interval change
-21. a few scattered calcifications again seen bilaterally when compared to the prior exam ( s ), there has been no significant interval change
-22. there is been no interval significant change in the multiple circumscribed nodules seen bilaterally as well as the coarse calcifications in both breasts
-23. there are extensive capsular calcifications on the left and mild capsular calcifications on the right , stable
-24. interval calcifications at the operative bed
-25. architectural distortion and dystrophic calcification in the left breast remain stable
-26. there is a stable benign nodule and benign - appearing calcifications
-27. these calcifications have increased since previous studies dating back to [ date ] although they have more significantly increased over the last several years
-28. there are adjacent stable benign - appearing large rodlike calcifications
-29. multiple bilateral scattered stable similar appearing nodules and multiple bilateral scattered secretory calcifications are considered benign
-30. a few isolated typically benign calcifications are seen bilaterally
-31. there are a few scattered calcifications , mildly progressive , but with a benign appearance
-32. no suspicious findings or significant changes are present as compared to the prior examination ( s ), the most recent dated [ date ] except that several small benign appearing calcifications are noted on the right new since the previous study
-33. a few benign - appearing calcifications again seen
-34. when compared to the prior exam ( s ), there has been progressive benign - appearing calcifications
-35. a large benign dystrophic calcification in the retroareolar right breast is present
-36. there are interval grouped calcifications at the operative bed
-37. a few small stable benign calcifications are present in both breasts
-38. there are a few scattered benign - appearing calcifications as well as vascular calcifications
-39. no suspicious findings or significant changes are present as compared to the prior examination ( s ), the most recent dated [ date ] except for the calcifications are more visualized on the current study
-40. the calcifications again seen
-41. there are a few calcifications in both breasts , mildly progressive but with a benign appearance
-42. there are bilateral stable scattered benign - appearing calcifications
-43. bilateral nodule stable benign appearing calcifications are seen in both breasts
-44. progressive grouped calcifications in the left breast
-45. benign secretory calcifications can be identified bilaterally
-46. there are a few bilateral stable benign - appearing calcifications
-47. there are a few scattered calcifications , mildly progressive but with a benign appearance
-48. secretory calcifications and calcified oil cysts are again scattered throughout both breasts
-49. small stable appearing left - sided calcifications are again noted
-50. multiple benign calcifications are present bilaterally without significant a biopsy clip in the left breast is again seen
-51. there are stable benign - appearing calcifications bilaterally
-52. bilateral benign - appearing stable oil cyst and calcifications are noted
-53. there are innumerable bilateral benign - appearing calcifications
-54. there are multiple scattered benign - appearing calcifications and a few scattered benign - appearing densities bilaterally
-55. there are mildly progressive bilateral calcifications which have a benign appearance
-56. several benign - appearing calcifications are noted on the right
-57. post - surgical changes are noted in the left breast and left axilla with clips and dystrophic calcifications noted
-58. tiny stable calcifications are noted on the left
-59. bilateral nodule stable benign appearing calcifications are seen in the left breast
-60. a few benign stable calcifications are present in both breasts
-61. numerous scattered benign type calcifications can be identified in all 4 quadrants bilaterally
-62. a few scattered calcifications are again seen bilaterally , mildly progressive but with a benign appearance
-63. a few scattered calcifications are again seen
-64. a few small stable benign - appearing calcifications are present in both breasts
-65. a few stable benign - appearing calcifications are noted
-66. stable nodular pattern with scattered coarse calcifications
-67. a few benign - appearing calcifications are noted bilaterally
-68. a few small stable and benign - appearing calcifications are present in both breasts
-69. there are scattered small calcifications , mildly progressive but with a benign appearance
-70. expected benign features of evolving fat necrosis are evident , including oral cyst formation and rare coarse calcification
-71. benign calcifications are present bilaterally without significant change
-72. benign retroareolar calcifications can again be identified bilaterally
-73. stable benign appearing left breast calcifications
-74. there are a few scattered calcifications , progressive , but with a benign appearance
-75. a benign stable dystrophic calcification is in the upper outer left breast at mid depth
-76. benign - appearing calcifications and nodules are present bilaterally
-77. there are a few scattered calcifications bilaterally which are mildly progressive but have a benign appearance
-78. bilateral scattered stable nodules and a few bilateral diffusely scattered punctate calcifications are again seen
-79. there are a few stable calcifications bilaterally
-80. bilateral benign / stable postsurgical change , and benign calcification
-81. a few benign - appearing calcifications are present
-82. small stable calcifications are noted bilaterally
-83. scattered benign type calcifications can be identified bilaterally
-84. there are scattered benign - appearing calcifications
-85. stable benign calcifications and nodularity is are once again noted bilaterally
-86. there are a few scattered calcifications in both breasts , mildly progressive but with a benign appearance
-87. there is a interval calcification in the operative bed on the left which has a dystrophic benign appearance
-88. a few benign - appearing calcifications are present bilaterally
-89. a calcified oil cyst is noted on the left
-90. stable small calcifications are noted bilaterally
-91. stable diffuse benign - appearing calcifications are present in the right breast
-92. there are stable benign - appearing large rodlike calcifications within both breasts
-93. a few small stable benign calcifications are present in the left breast
-94. postoperative changes and benign calcifications are stable in both breasts
-95. a few calcifications again seen bilaterally , mildly progressive but with a benign appearance
-96. there are several scattered benign - appearing calcifications
-97. left breast progressive calcifications as described above
-98. a couple of benign calcifications are present bilaterally
-99. stable benign calcifications are present on the left
-100. bilateral nodule stable benign appearing calcifications are seen in the right breast
-
-    
-    Task: 
-    Ignore '___' as they are due to anonymization.
-    Consider the consistent attributes present in the descriptions of correctly classified and misclassified samples regarding the positive cancer patients. Formulate hypotheses based on these attributes. Attributes are all the concepts (e.g, explicit or implicit anatomies, observations, any symptom of change related to the disease, demography related information or any concept leading to potential bias) in the sentences other than the class label (Cancer in this case). Assess how these characteristics might be influencing the classifier's performance. Your response should contain only the list of top hypothesis, nothing else. For the response, you should be the following python dictionary template, no extra sentence:
-    
-        hypothesis_dict = {
-        "H1": "The classifier is making mistake  as it is biased toward <attribute>",
-        "H2": "The classifier is making mistake as it is biased toward <attribute>",
-        "H3": "The classifier is making mistake as it is biased toward <attribute>",
-        ...
-        }
-    
-    
-    To effectively test Hypothesis 1 (H1) using the CLIP language encoder, you need to create prompts that explicitly validate H1. These prompts will help to generate text embeddings that capture the essence of the hypothesis, which can be used to compute similarity with the image embeddings from the dataset. The goal is to see if the images where the model makes mistakes are those that aligns with H1 or violates H1. The prompts are python list. Remember, your focus is only the class label "Cancer" (i.e, positive cancer cases)
-    
-    Do this for all the hypothesis. Your final response should follow the following list of dictionaries, nothing else:
-    
-    
-    prompt_dict = {
-        "H1_<attribute>": [List of prompts],
-        "H2_<attribute>": [List of prompts]
-        ...
-    }
-
-    
-    
-    Each attribute hypothesis should contain 5 prompts.
-    
-    So final response should follow the below format strictly (nothing else, no extra sentence): 
-    ```python
-        hypothesis_dict 
-        prompt_dict 
-    ```
-    

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/test_additional_info.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c67ea08cd95de55978d206db36f0612c6bbafde57f57dbf6033f0751b322ac3d
-size 197172

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/test_classifier_embeddings.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:82f8c49c3c558e77ef6ab4f3299e946da61224b054511fa7c242821f495c641a
-size 33546368

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/test_clip_embeddings.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ba704bcaa0c57a6528de4dec663a2931fd0b536214833807b52e517459bdcbdb
-size 8386688

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/train_additional_info.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:933db7f77d528028c61b9b84ca5506e26ddb1c4dddc26711fc531bdac6b4f412
-size 551343

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/train_classifier_embeddings.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:95bec61229256c99f4218051de9c4e3ac7eb77f5b05552ccb384914a1df81169
-size 93986944

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/train_clip_embeddings.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3db38e78923889e5cb1799f316773d02f343bb669c3ef41d5d42838b8f996414
-size 23496832

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/valid_additional_info.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:f033118b1e1219371f4731c9f96e7b9608f9a75feb927f7bc42af55849da9ff5
-size 236676

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/valid_classifier_embeddings.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a992d502fcc472eb3190da5ce4ca3d041e3b55af1fae6e7d7a97b8691404162e
-size 40288384

ViNDr/fold0/clip_img_encoder_tf_efficientnet_b5_ns-detect/valid_clip_embeddings.npy

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7cb5c4c7f96e0ab1074312dca301b4fb763b065098817051ab760afe5b34da8c
-size 10072192

ViNDr/fold0/efficientnetb5_seed_10_fold0_best_aucroc_ver084.pth

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ba8e02a098e94f24d662a4bd657a9a624e9dde8c558d7778ac87cc043d6111ed
-size 114419542

ViNDr/fold0/seed_10_train_configs.pkl

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a11eb6715fc72840c14dabbcc1632a0fa7db2a6b7704289c7076b5879bcfd96a
-size 1545