iapytoo 0.0.6.dev0__tar.gz → 0.0.6.dev2__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.

Potentially problematic release.


This version of iapytoo might be problematic. Click here for more details.

Files changed (74) hide show
  1. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/PKG-INFO +1 -1
  2. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/config_infer.yml +1 -1
  3. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/config_mnist.yml +1 -1
  4. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/config_wgan.yml +1 -0
  5. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/mlflow_infer.py +8 -6
  6. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/mnist.py +57 -38
  7. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/mnist_infer.py +12 -11
  8. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/wgan_mnist.py +6 -6
  9. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/_version.py +2 -2
  10. iapytoo-0.0.6.dev2/iapytoo/metrics/collection.py +39 -0
  11. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/metrics/metric.py +3 -3
  12. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/predictions/__init__.py +24 -8
  13. iapytoo-0.0.6.dev2/iapytoo/predictions/predictors.py +13 -0
  14. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/factories.py +170 -100
  15. iapytoo-0.0.6.dev2/iapytoo/train/inference.py +175 -0
  16. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/logger.py +2 -15
  17. iapytoo-0.0.6.dev2/iapytoo/train/mlflow_model.py +88 -0
  18. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/training.py +14 -35
  19. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/valuator.py +18 -0
  20. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/wgan.py +14 -31
  21. iapytoo-0.0.6.dev2/iapytoo/utils/__init__.py +0 -0
  22. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/model_config.py +3 -1
  23. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo.egg-info/PKG-INFO +1 -1
  24. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo.egg-info/SOURCES.txt +1 -0
  25. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/tests/test_metrics.py +1 -1
  26. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/tests/test_model.py +3 -3
  27. iapytoo-0.0.6.dev0/iapytoo/metrics/__init__.py +0 -82
  28. iapytoo-0.0.6.dev0/iapytoo/predictions/predictors.py +0 -33
  29. iapytoo-0.0.6.dev0/iapytoo/train/inference.py +0 -93
  30. iapytoo-0.0.6.dev0/iapytoo/train/mlflow_model.py +0 -80
  31. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/.github/workflows/python-app.yml +0 -0
  32. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/.gitignore +0 -0
  33. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/LICENSE +0 -0
  34. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/MANIFEST.in +0 -0
  35. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/README.md +0 -0
  36. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/dockers/Dockerfile +0 -0
  37. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/dockers/Readme.md +0 -0
  38. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/dockers/docker-compose.yml +0 -0
  39. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/dockers/requirements.txt +0 -0
  40. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/dockers/run-python.sh +0 -0
  41. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/examples/zipped_runs/selected_runs.zip +0 -0
  42. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/__init__.py +0 -0
  43. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/dataset/__init__.py +0 -0
  44. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/dataset/scaling.py +0 -0
  45. {iapytoo-0.0.6.dev0/iapytoo/train → iapytoo-0.0.6.dev2/iapytoo/metrics}/__init__.py +0 -0
  46. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/metrics/distrib.py +0 -0
  47. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/metrics/predefined.py +0 -0
  48. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/predictions/feature.py +0 -0
  49. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/predictions/plotters.py +0 -0
  50. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/predictions/types.py +0 -0
  51. {iapytoo-0.0.6.dev0/iapytoo/utils → iapytoo-0.0.6.dev2/iapytoo/train}/__init__.py +0 -0
  52. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/checkpoint.py +0 -0
  53. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/context.py +0 -0
  54. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/train/loss.py +0 -0
  55. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/arguments.py +0 -0
  56. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/cf/config.json +0 -0
  57. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/cf/config.yml +0 -0
  58. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/cf/config_test.json +0 -0
  59. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/config.py +0 -0
  60. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/display.py +0 -0
  61. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/iterative_mean.py +0 -0
  62. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/singleton.py +0 -0
  63. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/time_utils.py +0 -0
  64. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo/utils/timer.py +0 -0
  65. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo.egg-info/dependency_links.txt +0 -0
  66. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/iapytoo.egg-info/top_level.txt +0 -0
  67. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/mlflow/relocate_mlruns.py +0 -0
  68. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/mlflow/zip_selected_runs.py +0 -0
  69. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/pyproject.toml +0 -0
  70. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/requirements.txt +0 -0
  71. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/setup.cfg +0 -0
  72. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/tests/test_config.py +0 -0
  73. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/tests/test_dataset.py +0 -0
  74. {iapytoo-0.0.6.dev0 → iapytoo-0.0.6.dev2}/tests/test_scaling.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: iapytoo
3
- Version: 0.0.6.dev0
3
+ Version: 0.0.6.dev2
4
4
  Summary: Un toolkit pour vite coder un modèle pytorch d'IA utilisant MLflow
5
5
  Author-email: Jean-François Lecomte <zwergon@free.fr>
6
6
  License: Apache-2.0
@@ -14,4 +14,4 @@ metrics:
14
14
  model:
15
15
  type: "mlflow"
16
16
  predictor: "max"
17
- run_id: "59606474825b41b68b536f80d0b1911e"
17
+ run_id: "5bfea07ccb91434fafe6fc1def3c2011"
@@ -14,7 +14,7 @@ training:
14
14
  optimizer: "adam"
15
15
  scheduler: "step"
16
16
  epochs: 2
17
- tqdm: false
17
+ #tqdm: false
18
18
 
19
19
 
20
20
  metrics:
@@ -21,6 +21,7 @@ training:
21
21
 
22
22
  model:
23
23
  type: "gan"
24
+ valuator: "gan"
24
25
  generator: generator
25
26
  discriminator: critic
26
27
  n_critic: 1
@@ -13,23 +13,25 @@ def get_model_input(idx=0):
13
13
  )
14
14
  # prend une image du dataset MNIST de train
15
15
  pic: Image = dataset1[idx][0]
16
+ expected = dataset1[idx][1]
16
17
  # cree le numpy array
17
18
  img_input = np.array(pic.getdata()).reshape(
18
- 1, pic.size[0], pic.size[1]).astype(np.float32)
19
+ 1, pic.size[0], pic.size[1]).astype(np.float64)
19
20
  img_input /= 255.
20
21
 
21
22
  # crée un batch avec une image [1, 1, 28, 28]
22
- return np.array([img_input])
23
+ return np.array([img_input]), expected
23
24
 
24
25
 
25
26
  if __name__ == "__main__":
26
- run_id = "a821dbc9a0974bb0815479634c0e845c"
27
+ import mlflow.pyfunc as mp
28
+ run_id = "5bfea07ccb91434fafe6fc1def3c2011"
27
29
 
28
30
  logged_model = f'runs:/{run_id}/model'
29
31
 
30
32
  # Load model as a PyFuncModel.
31
- loaded_model = mlflow.pyfunc.load_model(logged_model)
33
+ loaded_model: mp.PyFuncModel = mlflow.pyfunc.load_model(logged_model)
32
34
 
33
- model_input = get_model_input()
35
+ model_input, expected = get_model_input()
34
36
  predicted = loaded_model.predict(model_input)
35
- print(predicted)
37
+ print(predicted[0], expected)
@@ -5,13 +5,14 @@ import torch.nn.functional as F
5
5
  from torch.utils.data import DataLoader
6
6
  from torchvision import datasets, transforms
7
7
  from torch.optim.lr_scheduler import LambdaLR
8
+ from PIL import Image
8
9
 
9
10
 
10
11
  from iapytoo.predictions.plotters import ConfusionPlotter
11
- from iapytoo.train.factories import Model, ModelFactory, SchedulerFactory, Scheduler
12
+ from iapytoo.train.factories import Model, Scheduler, Factory
12
13
  from iapytoo.utils.config import ConfigFactory, Config
13
14
  from iapytoo.train.training import Training
14
- from iapytoo.train.mlflow_model import Transform
15
+ from iapytoo.train.mlflow_model import MlflowTransform, IMlfowModelProvider
15
16
 
16
17
 
17
18
  from mlflow.types.schema import TensorSpec, Schema
@@ -58,32 +59,47 @@ class MnistScheduler(Scheduler):
58
59
  self.lr_scheduler = LambdaLR(optimizer, lr_lambda=lr_lambda)
59
60
 
60
61
 
61
- class MnistTraining(Training):
62
- def __init__(self, config: Config) -> None:
63
- super().__init__(config)
64
- self.predictions.add_plotter(ConfusionPlotter())
62
+ class MnistTransform(MlflowTransform):
65
63
 
66
-
67
- class MnistTransform(Transform):
68
-
69
- def __init__(self, dataset):
64
+ def __init__(self) -> None:
70
65
  super().__init__(
71
- train_transform=transforms.Compose(
66
+ transform=transforms.Compose(
72
67
  [
73
68
  transforms.ToTensor(),
74
69
  transforms.Normalize((0.1307,), (0.3081,))
75
70
  ]
76
- ),
77
- dataset=dataset
71
+ )
78
72
  )
79
73
 
80
74
  # override
81
- def set_signature(self, dataset):
82
- img, Y = dataset[0]
83
- X = Transform.img_to_numpy(img)
75
+ def __call__(self, model_input, *args, **kwds):
76
+
77
+ images_pil = []
78
+ for i in range(model_input.shape[0]):
79
+ # Extraire l'image (en supprimant la dimension du canal si nécessaire)
80
+ # Supposons que l'image est en niveaux de gris
81
+ img_array = model_input[i, 0, :, :]
82
+
83
+ # Normaliser l'image si nécessaire (par exemple, si les valeurs sont entre 0 et 1)
84
+ img_array = (img_array * 255).astype(np.uint8)
85
+
86
+ # Créer une image PIL
87
+ # 'L' pour les niveaux de gris
88
+ img_pil = Image.fromarray(img_array, 'L')
89
+ images_pil.append(img_pil)
90
+
91
+ model_input_tensor = torch.stack(
92
+ [self.transform(img) for img in images_pil])
93
+
94
+ return model_input_tensor
95
+
96
+
97
+ class MnistMlfowModel(IMlfowModelProvider):
98
+
99
+ def __init__(self):
100
+ X = np.random.rand(1, 1, 28, 28)
101
+ Y = np.zeros(shape=(1,), dtype=np.int64)
84
102
  # add batch size
85
- X = np.expand_dims(X, axis=0)
86
- Y = np.expand_dims(Y, axis=0)
87
103
  x_shape = list(X.shape)
88
104
  x_shape[0] = -1
89
105
  y_shape = list(Y.shape)
@@ -97,26 +113,35 @@ class MnistTransform(Transform):
97
113
  self.signature = ModelSignature(
98
114
  inputs=input_schema, outputs=output_schema)
99
115
  self.input_example = X
116
+ self.transform: MlflowTransform = MnistTransform()
100
117
 
101
118
  # override
119
+ def get_transform(self) -> MlflowTransform:
120
+ return self.transform
102
121
 
103
- def transform(self, context, model_input, params=None):
122
+ # override
123
+ def get_signature(self) -> ModelSignature:
124
+ return self.signature
104
125
 
105
- model_input_transformed = [
106
- self.infer_transform(img) for img in model_input]
107
- print(model_input_transformed[0].permute(1, 0, 2).shape)
126
+ # override
127
+ def get_input_example(self) -> np.array:
128
+ return self.input_example
108
129
 
109
- model_input_tensor = torch.stack(
110
- [img.permute(1, 0, 2) for img in model_input_transformed])
111
130
 
112
- return model_input_tensor
131
+ class MnistTraining(Training):
132
+
133
+ def __init__(self, config: Config) -> None:
134
+ super().__init__(config)
135
+ self.predictions.add_plotter(ConfusionPlotter())
136
+ self.mlflow_model_provider: IMlfowModelProvider = MnistMlfowModel()
113
137
 
114
138
 
115
139
  if __name__ == "__main__":
116
140
  from iapytoo.utils.arguments import parse_args
117
141
 
118
- ModelFactory().register_model("mnist", MnistModel)
119
- SchedulerFactory().register_scheduler("mnist", MnistScheduler)
142
+ factory = Factory()
143
+ factory.register_model("mnist", MnistModel)
144
+ factory.register_scheduler("mnist", MnistScheduler)
120
145
 
121
146
  args = parse_args()
122
147
 
@@ -129,24 +154,19 @@ if __name__ == "__main__":
129
154
 
130
155
  Training.seed(config)
131
156
 
132
- signature_dataset = datasets.MNIST(
133
- config.dataset.path,
134
- train=False
135
- )
136
- transform = MnistTransform(signature_dataset)
137
-
138
157
  training = MnistTraining(config)
158
+ mflow_transform: MlflowTransform = training.get_transform()
139
159
  train_dataset = datasets.MNIST(
140
160
  config.dataset.path,
141
161
  train=True,
142
162
  download=True,
143
- transform=transform.train_transform
163
+ transform=mflow_transform.transform
144
164
  )
145
165
 
146
- train_dataset = datasets.MNIST(
166
+ test_dataset = datasets.MNIST(
147
167
  config.dataset.path,
148
168
  train=False,
149
- transform=transform.infer_transform
169
+ transform=mflow_transform.transform
150
170
  )
151
171
 
152
172
  train_loader = DataLoader(
@@ -155,13 +175,12 @@ if __name__ == "__main__":
155
175
  num_workers=config.dataset.num_workers
156
176
  )
157
177
  test_loader = DataLoader(
158
- train_dataset, batch_size=config.dataset.batch_size,
178
+ test_dataset, batch_size=config.dataset.batch_size,
159
179
  num_workers=config.dataset.num_workers
160
180
  )
161
181
 
162
182
  training.fit(
163
183
  train_loader=train_loader,
164
184
  valid_loader=test_loader,
165
- transform=transform,
166
185
  run_id=args.run_id
167
186
  )
@@ -1,10 +1,10 @@
1
- import torch
2
- from torchvision import datasets, transforms
3
-
1
+ from torchvision import datasets
2
+ from torch.utils.data import DataLoader
4
3
  from iapytoo.predictions.plotters import ConfusionPlotter
5
- from iapytoo.utils.config import Config, ConfigFactory
4
+ from iapytoo.utils.config import Config, ConfigFactory, DatasetConfig
6
5
 
7
6
  from iapytoo.train.inference import MLFlowInference
7
+ from iapytoo.train.mlflow_model import MlflowTransform
8
8
 
9
9
 
10
10
  class MnistInference(MLFlowInference):
@@ -22,18 +22,19 @@ if __name__ == "__main__":
22
22
 
23
23
  inference = MnistInference(config)
24
24
 
25
- transform = transforms.Compose(
26
- [transforms.ToTensor(), transforms.Normalize((0.1307,), (0.3081,))]
27
- )
25
+ transform: MlflowTransform = inference.get_transform()
26
+
27
+ dataset_config: DatasetConfig = config.dataset
28
28
 
29
29
  test_dataset = datasets.MNIST(
30
- config.dataset.path,
30
+ dataset_config.path,
31
31
  train=False,
32
- transform=transform
32
+ transform=transform.transform
33
33
  )
34
34
 
35
- test_loader = torch.utils.data.DataLoader(
36
- test_dataset, batch_size=config.dataset.batch_size,
35
+ test_loader = DataLoader(
36
+ test_dataset,
37
+ batch_size=dataset_config.batch_size,
37
38
  num_workers=2
38
39
  )
39
40
 
@@ -8,8 +8,8 @@ import torchvision.transforms as transforms
8
8
 
9
9
 
10
10
  from iapytoo.train.wgan import WGAN
11
- from iapytoo.utils.config import Config
12
- from iapytoo.train.factories import Model, ModelFactory
11
+ from iapytoo.utils.config import ConfigFactory, Config
12
+ from iapytoo.train.factories import Model, Factory
13
13
  from iapytoo.predictions.plotters import Fake2DPlotter
14
14
  from iapytoo.train.factories import WeightInitiator
15
15
 
@@ -161,7 +161,7 @@ class LatentDataset(Dataset):
161
161
 
162
162
 
163
163
  if __name__ == "__main__":
164
- config = Config.create_from_yaml(os.path.join(
164
+ config = ConfigFactory.from_yaml(os.path.join(
165
165
  os.path.dirname(__file__), "config_wgan.yml"))
166
166
 
167
167
  transform = transforms.Compose(
@@ -188,9 +188,9 @@ if __name__ == "__main__":
188
188
  item = next(iter(valid_loader))
189
189
  print(item.shape)
190
190
 
191
- model_factory = ModelFactory()
192
- model_factory.register_model("generator", Generator)
193
- model_factory.register_model("critic", Discriminator)
191
+ factory = Factory()
192
+ factory.register_model("generator", Generator)
193
+ factory.register_model("critic", Discriminator)
194
194
 
195
195
  wgan = WGAN(config)
196
196
  wgan.predictions.add_plotter(Fake2DPlotter())
@@ -17,5 +17,5 @@ __version__: str
17
17
  __version_tuple__: VERSION_TUPLE
18
18
  version_tuple: VERSION_TUPLE
19
19
 
20
- __version__ = version = '0.0.6.dev0'
21
- __version_tuple__ = version_tuple = (0, 0, 6, 'dev0')
20
+ __version__ = version = '0.0.6.dev2'
21
+ __version_tuple__ = version_tuple = (0, 0, 6, 'dev2')
@@ -0,0 +1,39 @@
1
+ from iapytoo.utils.config import Config
2
+ from iapytoo.metrics.metric import Metric
3
+
4
+
5
+ class MetricsCollection(Metric):
6
+ def __init__(self, tag: str, metric_names: list[str], config: Config):
7
+ from iapytoo.train.factories import Factory, MetricError
8
+ super().__init__(tag, config)
9
+ self.metrics = {}
10
+ try:
11
+ factory = Factory()
12
+ for n in metric_names:
13
+ self.metrics[f"{tag}_{n}"] = factory.create_metric(n, config)
14
+ except MetricError as er:
15
+ print(f"Unable to create metric : {er}")
16
+
17
+ def to(self, device):
18
+ for m in self.metrics.values():
19
+ m.to(device)
20
+
21
+ # overwrite
22
+ def update(self, Y_hat, Y):
23
+ for m in self.metrics.values():
24
+ m.update(Y_hat, Y)
25
+
26
+ # overwrite
27
+ def compute(self):
28
+ self.results = {}
29
+ for k, m in self.metrics.items():
30
+ results = m.compute()
31
+ for name, result in results.items():
32
+ # prepend the name of the collection to all inner results : validation_accuracy for example
33
+ self.results[f"{self.name}_{name}"] = result
34
+ return self.results
35
+
36
+ # overwrite
37
+ def reset(self) -> None:
38
+ for m in self.metrics.values():
39
+ m.reset()
@@ -1,15 +1,15 @@
1
1
  import torch
2
2
 
3
3
  from iapytoo.utils.config import Config
4
- from iapytoo.predictions.predictors import PredictorFactory
5
4
 
6
5
 
7
6
  class Metric:
8
7
  def __init__(self, name, config: Config, with_target=True):
8
+ from iapytoo.train.factories import Factory
9
9
  self.with_target = with_target
10
10
  self.name = name
11
11
  self.config = config
12
- self.predictor = PredictorFactory().create_predictor(config)
12
+ self.predictor = Factory().create_predictor(config)
13
13
 
14
14
  # First dimension is for the concatenation.
15
15
  self.outputs = torch.zeros(size=(0,))
@@ -23,7 +23,7 @@ class Metric:
23
23
 
24
24
  @property
25
25
  def predicted(self):
26
- return self.predictor(self.outputs) if self.predictor else self.outputs
26
+ return self.predictor(self.outputs)
27
27
 
28
28
  def to(self, device):
29
29
  self.outputs = self.outputs.to(device)
@@ -1,22 +1,35 @@
1
+ from __future__ import annotations
1
2
  import torch
2
3
 
3
- from iapytoo.utils.config import Config
4
4
  from .types import PredictionType
5
5
  from .plotters import CollectionPlotters
6
- from .predictors import PredictorFactory
7
- from iapytoo.train.valuator import Valuator
6
+
7
+ from typing import TYPE_CHECKING
8
+
9
+ if TYPE_CHECKING:
10
+ from iapytoo.train.inference import Inference
8
11
 
9
12
 
10
13
  class Predictions:
11
- def __init__(self, config: Config):
14
+ def __init__(self, inference: Inference):
12
15
  self.outputs = None # dimension is assigned in compute
13
16
  self.actual = None
14
17
 
15
- self.predictor = PredictorFactory().create_predictor(config)
18
+ self.inference = inference
16
19
 
17
20
  self.prediction_plotter: CollectionPlotters = CollectionPlotters()
18
21
  self.prediction_plotter.connect(self)
19
22
 
23
+ @property
24
+ def valuator(self):
25
+ assert self.inference is not None, "no inference for this prediction"
26
+ return self.inference.valuator
27
+
28
+ @property
29
+ def predictor(self):
30
+ assert self.inference is not None, "no inference for this prediction"
31
+ return self.inference.predictor
32
+
20
33
  def __len__(self):
21
34
  """return the number of plotters used by the predictions"""
22
35
  return len(self.prediction_plotter)
@@ -24,12 +37,15 @@ class Predictions:
24
37
  def add_plotter(self, prediction_plotter):
25
38
  self.prediction_plotter.add(prediction_plotter)
26
39
 
27
- def compute(self, loader, valuator: Valuator):
40
+ def compute(self, loader):
41
+
42
+ assert self.valuator is not None, "no valuator"
43
+ assert self.predictor is not None, "no predictor"
28
44
 
29
45
  self.outputs = torch.zeros(size=(0,))
30
46
  self.actual = torch.zeros(size=(0,))
31
47
 
32
- for outputs, actual in valuator.evaluate_loader(loader):
48
+ for outputs, actual in self.valuator.evaluate_loader(loader):
33
49
  self.outputs = torch.cat((self.outputs, outputs), dim=0)
34
50
 
35
51
  if actual is not None:
@@ -37,7 +53,7 @@ class Predictions:
37
53
 
38
54
  def tensor(self, type: PredictionType = PredictionType.PREDICTED):
39
55
  if type == PredictionType.PREDICTED:
40
- return self.predictor(self.outputs) if self.predictor else self.outputs
56
+ return self.predictor(self.outputs)
41
57
  elif type == PredictionType.ACTUAL:
42
58
  return self.actual
43
59
  else:
@@ -0,0 +1,13 @@
1
+ import torch
2
+
3
+
4
+ class Predictor:
5
+ def __call__(self, outputs: torch.Tensor) -> torch.Tensor:
6
+ # By default, do nothing outputs of the model are predictions.
7
+ return outputs
8
+
9
+
10
+ class MaxPredictor(Predictor):
11
+ def __call__(self, outputs: torch.Tensor) -> torch.Tensor:
12
+ _, preds = torch.max(outputs.data, 1)
13
+ return preds