kcai-data-sampling 0.0.3__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (15) hide show
  1. kcai_data_sampling-0.0.3/PKG-INFO +83 -0
  2. kcai_data_sampling-0.0.3/README.md +51 -0
  3. kcai_data_sampling-0.0.3/pyproject.toml +31 -0
  4. kcai_data_sampling-0.0.3/requirements.txt +25 -0
  5. kcai_data_sampling-0.0.3/setup.cfg +4 -0
  6. kcai_data_sampling-0.0.3/src/kcai_data_sampling/__init__.py +9 -0
  7. kcai_data_sampling-0.0.3/src/kcai_data_sampling/adversarial_run.py +524 -0
  8. kcai_data_sampling-0.0.3/src/kcai_data_sampling/augment_run.py +807 -0
  9. kcai_data_sampling-0.0.3/src/kcai_data_sampling/kc_inference.py +357 -0
  10. kcai_data_sampling-0.0.3/src/kcai_data_sampling/run_augment_pipeline.py +658 -0
  11. kcai_data_sampling-0.0.3/src/kcai_data_sampling.egg-info/PKG-INFO +83 -0
  12. kcai_data_sampling-0.0.3/src/kcai_data_sampling.egg-info/SOURCES.txt +13 -0
  13. kcai_data_sampling-0.0.3/src/kcai_data_sampling.egg-info/dependency_links.txt +1 -0
  14. kcai_data_sampling-0.0.3/src/kcai_data_sampling.egg-info/requires.txt +23 -0
  15. kcai_data_sampling-0.0.3/src/kcai_data_sampling.egg-info/top_level.txt +1 -0
kcai_data_sampling-0.0.3/PKG-INFO ADDED
@@ -0,0 +1,83 @@
1
+ Metadata-Version: 2.4
2
+ Name: kcai-data-sampling
3
+ Version: 0.0.3
4
+ Summary: Samples generation utilities managed with uv
5
+ Author-email: Safenai <support@safenai.io>, IRT SystemX <support@irt-systemx.fr>
6
+ License-Expression: Apache-2.0
7
+ Requires-Python: >=3.11
8
+ Description-Content-Type: text/markdown
9
+ Requires-Dist: pyarrow>=10.0.0
10
+ Requires-Dist: numpy>=1.20.0
11
+ Requires-Dist: pillow>=8.0.0
12
+ Requires-Dist: scikit-learn>=1.0.0
13
+ Requires-Dist: torch>=1.10.0
14
+ Requires-Dist: ipykernel>=6.29.5
15
+ Requires-Dist: torchvision>=0.20.1
16
+ Requires-Dist: torch==2.5.1
17
+ Requires-Dist: torchvision==0.20.1
18
+ Requires-Dist: captum==0.7.0
19
+ Requires-Dist: torch_pruning==1.3.7
20
+ Requires-Dist: tqdm==4.66.5
21
+ Requires-Dist: click==8.1.7
22
+ Requires-Dist: pyyaml
23
+ Requires-Dist: numpy<2
24
+ Requires-Dist: neural-compressor==3.1.1
25
+ Requires-Dist: thop
26
+ Requires-Dist: tabulate
27
+ Requires-Dist: pyarrow==17.0.0
28
+ Requires-Dist: albumentations>=2.0.8
29
+ Requires-Dist: autoattack==0.5.5
30
+ Requires-Dist: cleverhans==4.1.0
31
+ Requires-Dist: boto3>=1.40.51
32
+
33
+ # KC-DATA-SAMPLING
34
+
35
+ ## Installation rapide
36
+
37
+ ```bash
38
+ uv venv .venv
39
+ source .venv/bin/activate
40
+ uv sync
41
+ ```
42
+
43
+ ## Lancement
44
+
45
+ ### 1. Génération d'images augmentées et adversariales
46
+
47
+ ```bash
48
+ # Images corrompues (augmentations)
49
+ python augment_run.py --config examples/augment_all.yml
50
+
51
+ # Images adversariales (FGSM, PGD, C&W, APGD)
52
+ python adversarial_run.py --config examples/adversarial_fgsm.yml
53
+ python adversarial_run.py --config examples/adversarial_all.yml
54
+ ```
55
+
56
+ **Note**: Ces scripts génèrent maintenant les images **sans inférence**. Les colonnes générées sont :
57
+ - `sample_uuid` (nouveau UUID unique)
58
+ - `parent_sample_uuid` (lien vers l'échantillon d'origine)
59
+ - `image_hash` (hash SHA256)
60
+ - `sample_id`, `sample_type`, `class_id`, `class_name`, `split`
61
+ - `transform_id`, `transform_name`, `transform_level`, `transform_params`
62
+ - `image_bytes`
63
+
64
+ ### 2. Inférence sur les images
65
+
66
+ ```bash
67
+ # Inférence de base (garde toutes les colonnes + ajoute prédictions)
68
+ python kc_inference.py --config examples/kc_inference.yaml
69
+
70
+ # mode in-place (onmodifie le fichier directement)
71
+ python kc_inference.py --config examples/kc_inference.yaml --in-place
72
+ ```
73
+
74
+ #### Options d'inférence disponibles :
75
+
76
+ **Modes de sortie** :
77
+ - `label_only` : Seulement le label prédit
78
+ - `label_prob` : Label + probabilité
79
+ - `top_k` : Label + prob + top-K labels et probs
80
+ - `all_probs` : Label + prob + toutes les probabilités
81
+ - `logits` : Label + prob + logits
82
+ - `extended` : Label + prob + all_probs + logits (recommandé)
83
+ - `full` : Tout (label + prob + top_k + all_probs + logits)
kcai_data_sampling-0.0.3/README.md ADDED
@@ -0,0 +1,51 @@
1
+ # KC-DATA-SAMPLING
2
+
3
+ ## Installation rapide
4
+
5
+ ```bash
6
+ uv venv .venv
7
+ source .venv/bin/activate
8
+ uv sync
9
+ ```
10
+
11
+ ## Lancement
12
+
13
+ ### 1. Génération d'images augmentées et adversariales
14
+
15
+ ```bash
16
+ # Images corrompues (augmentations)
17
+ python augment_run.py --config examples/augment_all.yml
18
+
19
+ # Images adversariales (FGSM, PGD, C&W, APGD)
20
+ python adversarial_run.py --config examples/adversarial_fgsm.yml
21
+ python adversarial_run.py --config examples/adversarial_all.yml
22
+ ```
23
+
24
+ **Note**: Ces scripts génèrent maintenant les images **sans inférence**. Les colonnes générées sont :
25
+ - `sample_uuid` (nouveau UUID unique)
26
+ - `parent_sample_uuid` (lien vers l'échantillon d'origine)
27
+ - `image_hash` (hash SHA256)
28
+ - `sample_id`, `sample_type`, `class_id`, `class_name`, `split`
29
+ - `transform_id`, `transform_name`, `transform_level`, `transform_params`
30
+ - `image_bytes`
31
+
32
+ ### 2. Inférence sur les images
33
+
34
+ ```bash
35
+ # Inférence de base (garde toutes les colonnes + ajoute prédictions)
36
+ python kc_inference.py --config examples/kc_inference.yaml
37
+
38
+ # mode in-place (onmodifie le fichier directement)
39
+ python kc_inference.py --config examples/kc_inference.yaml --in-place
40
+ ```
41
+
42
+ #### Options d'inférence disponibles :
43
+
44
+ **Modes de sortie** :
45
+ - `label_only` : Seulement le label prédit
46
+ - `label_prob` : Label + probabilité
47
+ - `top_k` : Label + prob + top-K labels et probs
48
+ - `all_probs` : Label + prob + toutes les probabilités
49
+ - `logits` : Label + prob + logits
50
+ - `extended` : Label + prob + all_probs + logits (recommandé)
51
+ - `full` : Tout (label + prob + top_k + all_probs + logits)
kcai_data_sampling-0.0.3/pyproject.toml ADDED
@@ -0,0 +1,31 @@
1
+ [build-system]
2
+ requires = ["setuptools>=80", "setuptools-scm[simple]>=8"]
3
+ build-backend = "setuptools.build_meta"
4
+
5
+ [project]
6
+ name = "kcai-data-sampling"
7
+ version = "0.0.3"
8
+ description = "Samples generation utilities managed with uv"
9
+ dynamic = ["dependencies"] #, "version"]
10
+ authors = [
11
+ {name = "Safenai", email="support@safenai.io"},
12
+ {name = "IRT SystemX", email="support@irt-systemx.fr"},
13
+ ]
14
+ readme = "README.md"
15
+ requires-python = ">=3.11"
16
+ license ="Apache-2.0"
17
+
18
+ [tool.pytest.ini_options]
19
+ addopts = "-q"
20
+
21
+ [tool.ruff]
22
+ line-length = 100
23
+
24
+
25
+ [tool.setuptools.dynamic]
26
+ dependencies = {file = ["requirements.txt"]}
27
+
28
+ [tool.setuptools_scm]
29
+ version_file = "src/kcai_data_sampling/_version_.py"
30
+ root= "../.."
31
+
kcai_data_sampling-0.0.3/requirements.txt ADDED
@@ -0,0 +1,25 @@
1
+ # Main dependencies for kc-dataset
2
+ pyarrow>=10.0.0
3
+ numpy>=1.20.0
4
+ pillow>=8.0.0
5
+ scikit-learn>=1.0.0
6
+ torch>=1.10.0
7
+ ipykernel>=6.29.5
8
+ torchvision>=0.20.1
9
+
10
+ torch==2.5.1
11
+ torchvision==0.20.1
12
+ captum==0.7.0
13
+ torch_pruning==1.3.7
14
+ tqdm==4.66.5
15
+ click==8.1.7
16
+ pyyaml
17
+ numpy<2
18
+ neural-compressor==3.1.1
19
+ thop
20
+ tabulate
21
+ pyarrow==17.0.0
22
+ albumentations>=2.0.8
23
+ autoattack==0.5.5
24
+ cleverhans==4.1.0
25
+ boto3>=1.40.51
kcai_data_sampling-0.0.3/setup.cfg ADDED
@@ -0,0 +1,4 @@
1
+ [egg_info]
2
+ tag_build =
3
+ tag_date = 0
4
+
kcai_data_sampling-0.0.3/src/kcai_data_sampling/__init__.py ADDED
@@ -0,0 +1,9 @@
1
+ __description__ = "Generating samples......"
2
+
3
+ from .augment_run import run_pipeline as RunAugmentPipeline
4
+ from .adversarial_run import run_pipeline as RunAdversarialPipeline
5
+
6
+ __all__ = [
7
+ "RunAugmentPipeline",
8
+ "RunAdversarialPipeline"
9
+ ]