synthyverse 0.2.6__tar.gz → 0.2.7__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.
- {synthyverse-0.2.6 → synthyverse-0.2.7}/PKG-INFO +1 -1
- {synthyverse-0.2.6 → synthyverse-0.2.7}/pyproject.toml +1 -1
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/benchmark/synthesis.py +24 -33
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/cdtd_generator/cdtd.py +21 -32
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/ctgan_generator/ct_gan.py +5 -19
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/ctgan_generator/synthesizer.py +11 -8
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/dgm_utils.py +14 -6
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/fd.py +3 -3
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/synthpop_generator/synthpop.py +78 -20
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabcascade_generator/tabcascade.py +27 -30
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabddpm_generator/tabddpm.py +22 -32
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabdiff_generator/tabdiff.py +21 -30
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabsyn_generator/tabsyn.py +21 -27
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tvae_generator/synthesizer.py +11 -8
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tvae_generator/tvae.py +5 -19
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/xgbddpm_generator/xgbddpm.py +87 -34
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/xgbdiffusion_generator/xgbdiffusion.py +13 -9
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse.egg-info/PKG-INFO +1 -1
- {synthyverse-0.2.6 → synthyverse-0.2.7}/LICENSE +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/LICENSES/Apache-2.0.txt +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/LICENSES/CTGAN-BSL-1.1.txt +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/LICENSES/MIT.txt +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/MANIFEST.in +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/NOTICE +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/README.md +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/THIRD_PARTY_NOTICES.md +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/logo/logo.png +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/setup.cfg +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/benchmark/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/base.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/distances.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/eval.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/fidelity.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/ml.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/preprocessing.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/privacy.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/evaluation/utility.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/_optional.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/arf_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/arf_generator/arf.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/arf_generator/model.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/arf_generator/utils.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/base.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/cdtd_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/cdtd_generator/layers.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/cdtd_generator/utils.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/ctgan_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/model.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/utils/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/utils/diffusion.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/utils/utils_diffusion.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/utils/writer.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/persistence.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/smote_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/smote_generator/smote.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/synthpop_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabcascade_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabcascade_generator/disttree.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabcascade_generator/encoder.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabcascade_generator/highres.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabcascade_generator/lowres.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabddpm_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabddpm_generator/ddpm_utils.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabddpm_generator/gaussian_multinomial_diffsuion.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabddpm_generator/modules.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabdiff_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabdiff_generator/diffusion.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabdiff_generator/modules.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabdiff_generator/noise_schedule.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabdiff_generator/transformer.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabsyn_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabsyn_generator/diffusion.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tabsyn_generator/vae.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/tvae_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/univariate_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/univariate_generator/univariate.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/xgbddpm_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/xgbdiffusion_generator/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/utils/__init__.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/utils/reproducibility.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/utils/utils.py +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse.egg-info/SOURCES.txt +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse.egg-info/dependency_links.txt +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse.egg-info/requires.txt +0 -0
- {synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse.egg-info/top_level.txt +0 -0
|
@@ -66,7 +66,7 @@ class TabularSynthesisBenchmark:
|
|
|
66
66
|
``"mean"``, ``"median"``, ``"most_frequent"``, and
|
|
67
67
|
``"missforest"``. Default: ``"drop"``.
|
|
68
68
|
monitor_memory (bool): Whether to record peak CPU and, when available,
|
|
69
|
-
CUDA memory
|
|
69
|
+
CUDA memory used by training. Default: False.
|
|
70
70
|
reuse_processors (bool): Whether to reuse saved preprocessing artifacts
|
|
71
71
|
for each train seed when available. Default: True.
|
|
72
72
|
max_eval_samples (int or None): Default maximum number of rows per real
|
|
@@ -654,21 +654,12 @@ class TabularSynthesisBenchmark:
|
|
|
654
654
|
)
|
|
655
655
|
n_train_syn = len(X_train_eval)
|
|
656
656
|
|
|
657
|
-
memory_monitor = None
|
|
658
657
|
sampling_start_time = perf_counter()
|
|
659
|
-
|
|
660
|
-
|
|
661
|
-
|
|
662
|
-
|
|
663
|
-
|
|
664
|
-
)
|
|
665
|
-
X_syn = processor.postprocess(X_syn)
|
|
666
|
-
else:
|
|
667
|
-
X_syn = generator.generate(
|
|
668
|
-
n_train_syn,
|
|
669
|
-
random_state=sampling_seed,
|
|
670
|
-
)
|
|
671
|
-
X_syn = processor.postprocess(X_syn)
|
|
658
|
+
X_syn = generator.generate(
|
|
659
|
+
n_train_syn,
|
|
660
|
+
random_state=sampling_seed,
|
|
661
|
+
)
|
|
662
|
+
X_syn = processor.postprocess(X_syn)
|
|
672
663
|
|
|
673
664
|
X_syn = X_syn.reset_index(drop=True)
|
|
674
665
|
sampling_time = perf_counter() - sampling_start_time
|
|
@@ -686,9 +677,6 @@ class TabularSynthesisBenchmark:
|
|
|
686
677
|
"set": set_index,
|
|
687
678
|
}
|
|
688
679
|
)
|
|
689
|
-
result_rows.extend(
|
|
690
|
-
memory_metric_rows("sampling", memory_monitor, train_seed, set_index)
|
|
691
|
-
)
|
|
692
680
|
result_rows.extend(
|
|
693
681
|
self._evaluate_synthetic_datasets(
|
|
694
682
|
X_train_eval=X_train_eval,
|
|
@@ -1081,22 +1069,22 @@ def training_timeout_rows(
|
|
|
1081
1069
|
|
|
1082
1070
|
|
|
1083
1071
|
class PeakMemoryMonitor:
|
|
1084
|
-
"""Sample
|
|
1072
|
+
"""Sample memory used above the pre-training baseline."""
|
|
1085
1073
|
|
|
1086
1074
|
def __init__(self, interval_seconds: float = 1.0):
|
|
1087
1075
|
self.interval_seconds = interval_seconds
|
|
1088
1076
|
self.peak_memory_mb: Optional[float] = None
|
|
1089
1077
|
self.peak_cuda_memory_mb: Optional[float] = None
|
|
1090
1078
|
self._baseline_rss_bytes: Optional[int] = None
|
|
1091
|
-
self.
|
|
1079
|
+
self._baseline_cuda_memory_bytes: Optional[int] = None
|
|
1092
1080
|
self._peak_rss_bytes: Optional[int] = None
|
|
1093
1081
|
self._stop_event = threading.Event()
|
|
1094
1082
|
self._thread: Optional[threading.Thread] = None
|
|
1095
1083
|
|
|
1096
1084
|
def __enter__(self) -> "PeakMemoryMonitor":
|
|
1097
1085
|
self._baseline_rss_bytes = _current_process_memory_bytes()
|
|
1098
|
-
self.
|
|
1099
|
-
self.
|
|
1086
|
+
self._peak_rss_bytes = self._baseline_rss_bytes
|
|
1087
|
+
self._baseline_cuda_memory_bytes = _current_cuda_memory_bytes()
|
|
1100
1088
|
_reset_cuda_peak_memory()
|
|
1101
1089
|
self._thread = threading.Thread(target=self._sample_until_stopped, daemon=True)
|
|
1102
1090
|
self._thread.start()
|
|
@@ -1109,10 +1097,13 @@ class PeakMemoryMonitor:
|
|
|
1109
1097
|
self._thread.join(timeout=1.0)
|
|
1110
1098
|
self._sample_once()
|
|
1111
1099
|
self.peak_memory_mb = _bytes_to_mib(
|
|
1112
|
-
|
|
1100
|
+
_memory_delta_bytes(self._peak_rss_bytes, self._baseline_rss_bytes)
|
|
1113
1101
|
)
|
|
1114
1102
|
self.peak_cuda_memory_mb = _bytes_to_mib(
|
|
1115
|
-
|
|
1103
|
+
_memory_delta_bytes(
|
|
1104
|
+
_cuda_peak_memory_bytes(),
|
|
1105
|
+
self._baseline_cuda_memory_bytes,
|
|
1106
|
+
)
|
|
1116
1107
|
)
|
|
1117
1108
|
|
|
1118
1109
|
def _sample_until_stopped(self) -> None:
|
|
@@ -1133,13 +1124,13 @@ def _bytes_to_mib(value: Optional[int]) -> Optional[float]:
|
|
|
1133
1124
|
return value / BYTES_PER_MIB
|
|
1134
1125
|
|
|
1135
1126
|
|
|
1136
|
-
def
|
|
1137
|
-
|
|
1138
|
-
|
|
1127
|
+
def _memory_delta_bytes(
|
|
1128
|
+
peak_bytes: Optional[int],
|
|
1129
|
+
baseline_bytes: Optional[int],
|
|
1139
1130
|
) -> Optional[int]:
|
|
1140
|
-
if
|
|
1131
|
+
if peak_bytes is None or baseline_bytes is None:
|
|
1141
1132
|
return None
|
|
1142
|
-
return max(
|
|
1133
|
+
return max(peak_bytes - baseline_bytes, 0)
|
|
1143
1134
|
|
|
1144
1135
|
|
|
1145
1136
|
def _current_process_memory_bytes() -> Optional[int]:
|
|
@@ -1256,26 +1247,26 @@ def _reset_cuda_peak_memory() -> None:
|
|
|
1256
1247
|
return
|
|
1257
1248
|
|
|
1258
1249
|
|
|
1259
|
-
def
|
|
1250
|
+
def _current_cuda_memory_bytes() -> Optional[int]:
|
|
1260
1251
|
try:
|
|
1261
1252
|
import torch
|
|
1262
1253
|
|
|
1263
1254
|
if not torch.cuda.is_available():
|
|
1264
1255
|
return None
|
|
1265
1256
|
torch.cuda.synchronize()
|
|
1266
|
-
return int(torch.cuda.
|
|
1257
|
+
return int(torch.cuda.memory_allocated())
|
|
1267
1258
|
except Exception:
|
|
1268
1259
|
return None
|
|
1269
1260
|
|
|
1270
1261
|
|
|
1271
|
-
def
|
|
1262
|
+
def _cuda_peak_memory_bytes() -> Optional[int]:
|
|
1272
1263
|
try:
|
|
1273
1264
|
import torch
|
|
1274
1265
|
|
|
1275
1266
|
if not torch.cuda.is_available():
|
|
1276
1267
|
return None
|
|
1277
1268
|
torch.cuda.synchronize()
|
|
1278
|
-
return int(torch.cuda.
|
|
1269
|
+
return int(torch.cuda.max_memory_allocated())
|
|
1279
1270
|
except Exception:
|
|
1280
1271
|
return None
|
|
1281
1272
|
|
|
@@ -12,7 +12,6 @@ from ..dgm_utils import (
|
|
|
12
12
|
FastTensorDataLoader,
|
|
13
13
|
QuantileStandardScaler,
|
|
14
14
|
clone_state_dict,
|
|
15
|
-
split_validation,
|
|
16
15
|
validate_c2st,
|
|
17
16
|
)
|
|
18
17
|
from .layers import MLP, MixedTypeDiffusion
|
|
@@ -54,9 +53,7 @@ class CDTDGenerator(BaseGenerator):
|
|
|
54
53
|
ema_decay (float): Exponential moving average decay. Default: 0.999.
|
|
55
54
|
log_steps (int): Steps between logging. Default: 100.
|
|
56
55
|
cap_train_time (float): Time limit in seconds for training. Default: None.
|
|
57
|
-
|
|
58
|
-
val_steps (int): Steps between validation. Default: 5000.
|
|
59
|
-
target_column (str): Name of the target column, potentially used for stratified validation splitting. Default: None.
|
|
56
|
+
val_steps (int): Steps between training-set C2ST validation. Set to <=0 to disable validation. Default: 5000.
|
|
60
57
|
|
|
61
58
|
Example:
|
|
62
59
|
>>> import pandas as pd
|
|
@@ -106,9 +103,7 @@ class CDTDGenerator(BaseGenerator):
|
|
|
106
103
|
random_state: int = 0,
|
|
107
104
|
full_determinism: bool = False,
|
|
108
105
|
cap_train_time: Optional[float] = None,
|
|
109
|
-
val_size: float = 0.0,
|
|
110
106
|
val_steps: int = 5000,
|
|
111
|
-
target_column: Optional[str] = None,
|
|
112
107
|
):
|
|
113
108
|
super().__init__(random_state=random_state, full_determinism=full_determinism)
|
|
114
109
|
self.cat_emb_dim = cat_emb_dim
|
|
@@ -132,9 +127,7 @@ class CDTDGenerator(BaseGenerator):
|
|
|
132
127
|
self.ema_decay = ema_decay
|
|
133
128
|
self.log_steps = log_steps
|
|
134
129
|
self.cap_train_time = cap_train_time
|
|
135
|
-
self.val_size = val_size
|
|
136
130
|
self.val_steps = val_steps
|
|
137
|
-
self.target_column = target_column
|
|
138
131
|
|
|
139
132
|
def _fit(self, X: pd.DataFrame, discrete_features: list):
|
|
140
133
|
self.discrete_features = discrete_features
|
|
@@ -143,23 +136,15 @@ class CDTDGenerator(BaseGenerator):
|
|
|
143
136
|
]
|
|
144
137
|
self.col_order = X.columns
|
|
145
138
|
|
|
146
|
-
|
|
139
|
+
X_train = X.copy()
|
|
147
140
|
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
self.
|
|
151
|
-
self.target_column,
|
|
152
|
-
self.discrete_features,
|
|
153
|
-
self.random_state,
|
|
141
|
+
self.quant_encoder = QuantileStandardScaler(X_train.shape[0], self.random_state)
|
|
142
|
+
X_train[self.numerical_features] = self.quant_encoder.fit_transform(
|
|
143
|
+
X_train[self.numerical_features].astype(float)
|
|
154
144
|
)
|
|
155
145
|
|
|
156
|
-
|
|
157
|
-
|
|
158
|
-
X[self.numerical_features].astype(float)
|
|
159
|
-
)
|
|
160
|
-
|
|
161
|
-
X_discrete = torch.tensor(X[self.discrete_features].to_numpy()).long()
|
|
162
|
-
X_numerical = torch.tensor(X[self.numerical_features].to_numpy()).float()
|
|
146
|
+
X_discrete = torch.tensor(X_train[self.discrete_features].to_numpy()).long()
|
|
147
|
+
X_numerical = torch.tensor(X_train[self.numerical_features].to_numpy()).float()
|
|
163
148
|
|
|
164
149
|
# --- build diffusion model ---
|
|
165
150
|
self.num_cat_features = X_discrete.shape[1]
|
|
@@ -245,9 +230,10 @@ class CDTDGenerator(BaseGenerator):
|
|
|
245
230
|
best_val_score = float("inf")
|
|
246
231
|
best_val_model = None
|
|
247
232
|
|
|
248
|
-
|
|
233
|
+
train_time = 0.0
|
|
249
234
|
with tqdm(initial=current_step, total=self.training_steps) as pbar:
|
|
250
235
|
while current_step < self.training_steps:
|
|
236
|
+
step_start_time = time.monotonic()
|
|
251
237
|
optimizer.zero_grad()
|
|
252
238
|
|
|
253
239
|
inputs = next(train_iter)
|
|
@@ -277,8 +263,16 @@ class CDTDGenerator(BaseGenerator):
|
|
|
277
263
|
for param_group in optimizer.param_groups:
|
|
278
264
|
param_group["lr"] = scheduler(current_step)
|
|
279
265
|
|
|
266
|
+
train_time += time.monotonic() - step_start_time
|
|
267
|
+
if (
|
|
268
|
+
self.cap_train_time is not None
|
|
269
|
+
and train_time > self.cap_train_time
|
|
270
|
+
):
|
|
271
|
+
print(f"Training timed out after {self.cap_train_time} seconds.")
|
|
272
|
+
break
|
|
273
|
+
|
|
280
274
|
if (
|
|
281
|
-
|
|
275
|
+
self.val_steps > 0
|
|
282
276
|
and current_step % self.val_steps == 0
|
|
283
277
|
and current_step > 0
|
|
284
278
|
):
|
|
@@ -286,7 +280,9 @@ class CDTDGenerator(BaseGenerator):
|
|
|
286
280
|
self.diff_model.eval()
|
|
287
281
|
ema_diff_model.store()
|
|
288
282
|
ema_diff_model.copy_to()
|
|
289
|
-
|
|
283
|
+
|
|
284
|
+
score = validate_c2st(self, X, random_state=self.random_state)
|
|
285
|
+
print(f"Validation score at step {current_step}: {score}")
|
|
290
286
|
|
|
291
287
|
if score < best_val_score:
|
|
292
288
|
best_val_score = score
|
|
@@ -301,13 +297,6 @@ class CDTDGenerator(BaseGenerator):
|
|
|
301
297
|
if stop_training:
|
|
302
298
|
break
|
|
303
299
|
|
|
304
|
-
if (
|
|
305
|
-
self.cap_train_time is not None
|
|
306
|
-
and (time.monotonic() - start_time) > self.cap_train_time
|
|
307
|
-
):
|
|
308
|
-
print(f"Training timed out after {self.cap_train_time} seconds.")
|
|
309
|
-
break
|
|
310
|
-
|
|
311
300
|
if best_val_model is None:
|
|
312
301
|
ema_diff_model.copy_to()
|
|
313
302
|
else:
|
|
@@ -7,7 +7,7 @@ import pandas as pd
|
|
|
7
7
|
|
|
8
8
|
from .._optional import require_ctgan
|
|
9
9
|
from ..base import BaseGenerator
|
|
10
|
-
from ..dgm_utils import clone_state_dict,
|
|
10
|
+
from ..dgm_utils import clone_state_dict, validate_c2st
|
|
11
11
|
from ...utils.utils import resolve_epochs_from_training_steps
|
|
12
12
|
|
|
13
13
|
|
|
@@ -41,9 +41,7 @@ class CTGANGenerator(BaseGenerator):
|
|
|
41
41
|
discriminator. Default: 10.
|
|
42
42
|
cuda (bool): Whether to use CUDA if available. Default: True.
|
|
43
43
|
cap_train_time (float): Time limit in seconds for training. Default: None.
|
|
44
|
-
|
|
45
|
-
val_steps (int): Epochs between validation, or training steps when ``training_steps`` is provided. Default: 50.
|
|
46
|
-
target_column (str): Name of the target column, potentially used for stratified validation splitting. Default: None.
|
|
44
|
+
val_steps (int): Epochs between training-set C2ST validation, or training steps when ``training_steps`` is provided. Set to <=0 to disable validation. Default: 50.
|
|
47
45
|
|
|
48
46
|
Example:
|
|
49
47
|
>>> import pandas as pd
|
|
@@ -85,9 +83,7 @@ class CTGANGenerator(BaseGenerator):
|
|
|
85
83
|
pac=10,
|
|
86
84
|
cuda=True,
|
|
87
85
|
cap_train_time: Optional[float] = None,
|
|
88
|
-
val_size: float = 0.0,
|
|
89
86
|
val_steps: int = 50,
|
|
90
|
-
target_column: Optional[str] = None,
|
|
91
87
|
random_state: int = 0,
|
|
92
88
|
full_determinism: bool = False,
|
|
93
89
|
):
|
|
@@ -114,21 +110,13 @@ class CTGANGenerator(BaseGenerator):
|
|
|
114
110
|
raise ValueError("pac must be >= 1.")
|
|
115
111
|
self.cuda = cuda
|
|
116
112
|
self.cap_train_time = cap_train_time
|
|
117
|
-
self.val_size = val_size
|
|
118
113
|
self.val_steps = val_steps
|
|
119
|
-
self.target_column = target_column
|
|
120
114
|
|
|
121
115
|
def _fit(self, X: pd.DataFrame, discrete_features: list):
|
|
122
116
|
from .synthesizer import CTGAN
|
|
123
117
|
|
|
124
118
|
self.discrete_features = list(discrete_features)
|
|
125
|
-
X
|
|
126
|
-
X,
|
|
127
|
-
self.val_size,
|
|
128
|
-
self.target_column,
|
|
129
|
-
self.discrete_features,
|
|
130
|
-
self.random_state,
|
|
131
|
-
)
|
|
119
|
+
X = X.copy()
|
|
132
120
|
|
|
133
121
|
self.batch_size = self._round_batch_size(self.batch_size, self.pac)
|
|
134
122
|
epochs = resolve_epochs_from_training_steps(
|
|
@@ -164,7 +152,7 @@ class CTGANGenerator(BaseGenerator):
|
|
|
164
152
|
def validate():
|
|
165
153
|
nonlocal best_val_score, best_val_model
|
|
166
154
|
self.model._generator.eval()
|
|
167
|
-
score = validate_c2st(self,
|
|
155
|
+
score = validate_c2st(self, X, random_state=self.random_state)
|
|
168
156
|
self.model._generator.train()
|
|
169
157
|
if score < best_val_score:
|
|
170
158
|
best_val_score = score
|
|
@@ -173,7 +161,7 @@ class CTGANGenerator(BaseGenerator):
|
|
|
173
161
|
return True
|
|
174
162
|
|
|
175
163
|
def validate_callback(step, epoch, epoch_end):
|
|
176
|
-
if
|
|
164
|
+
if self.val_steps <= 0:
|
|
177
165
|
return False
|
|
178
166
|
if self.training_steps is not None:
|
|
179
167
|
return (
|
|
@@ -220,9 +208,7 @@ class CTGANGenerator(BaseGenerator):
|
|
|
220
208
|
"pac": self.pac,
|
|
221
209
|
"cuda": self.cuda,
|
|
222
210
|
"cap_train_time": self.cap_train_time,
|
|
223
|
-
"val_size": self.val_size,
|
|
224
211
|
"val_steps": self.val_steps,
|
|
225
|
-
"target_column": self.target_column,
|
|
226
212
|
"discrete_features": getattr(self, "discrete_features", None),
|
|
227
213
|
}
|
|
228
214
|
|
{synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/ctgan_generator/synthesizer.py
RENAMED
|
@@ -424,12 +424,13 @@ class CTGAN(BaseSynthesizer):
|
|
|
424
424
|
epoch_iterator.set_description(description.format(gen=0, dis=0))
|
|
425
425
|
|
|
426
426
|
steps_per_epoch = max(len(train_data) // self._batch_size, 1)
|
|
427
|
-
|
|
427
|
+
train_time = 0.0
|
|
428
428
|
step = 0
|
|
429
429
|
timed_out = False
|
|
430
430
|
stop_training = False
|
|
431
431
|
for i in epoch_iterator:
|
|
432
432
|
for id_ in range(steps_per_epoch):
|
|
433
|
+
step_start_time = time.monotonic()
|
|
433
434
|
for n in range(self._discriminator_steps):
|
|
434
435
|
fakez = torch.normal(mean=mean, std=std)
|
|
435
436
|
|
|
@@ -507,18 +508,19 @@ class CTGAN(BaseSynthesizer):
|
|
|
507
508
|
loss_g.backward()
|
|
508
509
|
optimizerG.step()
|
|
509
510
|
step += 1
|
|
510
|
-
|
|
511
|
-
step, i + 1, False
|
|
512
|
-
):
|
|
513
|
-
stop_training = True
|
|
514
|
-
break
|
|
511
|
+
train_time += time.monotonic() - step_start_time
|
|
515
512
|
if (
|
|
516
513
|
self.cap_train_time is not None
|
|
517
|
-
and
|
|
514
|
+
and train_time > self.cap_train_time
|
|
518
515
|
):
|
|
519
516
|
print(f"Training timed out after {self.cap_train_time} seconds.")
|
|
520
517
|
timed_out = True
|
|
521
518
|
break
|
|
519
|
+
if validate_callback is not None and validate_callback(
|
|
520
|
+
step, i + 1, False
|
|
521
|
+
):
|
|
522
|
+
stop_training = True
|
|
523
|
+
break
|
|
522
524
|
|
|
523
525
|
generator_loss = loss_g.detach().cpu().item()
|
|
524
526
|
discriminator_loss = loss_d.detach().cpu().item()
|
|
@@ -542,7 +544,8 @@ class CTGAN(BaseSynthesizer):
|
|
|
542
544
|
description.format(gen=generator_loss, dis=discriminator_loss)
|
|
543
545
|
)
|
|
544
546
|
if (
|
|
545
|
-
not
|
|
547
|
+
not timed_out
|
|
548
|
+
and not stop_training
|
|
546
549
|
and validate_callback is not None
|
|
547
550
|
and validate_callback(step, i + 1, True)
|
|
548
551
|
):
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
import random
|
|
2
2
|
from contextlib import contextmanager
|
|
3
|
-
|
|
3
|
+
from math import ceil
|
|
4
4
|
import numpy as np
|
|
5
5
|
import torch
|
|
6
6
|
import torch.nn as nn
|
|
@@ -30,16 +30,22 @@ def preserve_rng_state(generator):
|
|
|
30
30
|
|
|
31
31
|
|
|
32
32
|
def validate_c2st(
|
|
33
|
-
self,
|
|
33
|
+
self,
|
|
34
|
+
X: pd.DataFrame,
|
|
35
|
+
nfolds: int = 3,
|
|
36
|
+
random_state: int = 0,
|
|
37
|
+
max_samples: int = 50_000,
|
|
34
38
|
):
|
|
35
|
-
|
|
39
|
+
n = min(len(X), max_samples)
|
|
40
|
+
n_sets = int(max(min(ceil(max_samples / len(X)), 10), 3))
|
|
36
41
|
with preserve_rng_state(self):
|
|
37
42
|
result = 0
|
|
38
43
|
for i in range(n_sets):
|
|
39
44
|
state = random_state + i + 1
|
|
40
|
-
|
|
45
|
+
x = X.sample(n, replace=False, ignore_index=True, random_state=state)
|
|
46
|
+
syn = self.generate(n, random_state=state)
|
|
41
47
|
score = ClassifierTwoSampleTest(nfold=nfolds, random_state=state).evaluate(
|
|
42
|
-
|
|
48
|
+
x, syn, discrete_features=self.discrete_features
|
|
43
49
|
)["c2st.auc"]
|
|
44
50
|
result += abs(score - 0.5)
|
|
45
51
|
return result / n_sets
|
|
@@ -162,7 +168,9 @@ class FastTensorDataLoader:
|
|
|
162
168
|
|
|
163
169
|
n_batches, remainder = divmod(self.dataset_len, self.batch_size)
|
|
164
170
|
self.n_batches = n_batches if drop_last else n_batches + (remainder > 0)
|
|
165
|
-
self.iter_len =
|
|
171
|
+
self.iter_len = (
|
|
172
|
+
self.n_batches * self.batch_size if drop_last else self.dataset_len
|
|
173
|
+
)
|
|
166
174
|
|
|
167
175
|
def __iter__(self):
|
|
168
176
|
self.indices = torch.randperm(self.dataset_len) if self.shuffle else None
|
{synthyverse-0.2.6 → synthyverse-0.2.7}/synthyverse/generators/forestdiffusion_generator/fd.py
RENAMED
|
@@ -29,15 +29,15 @@ class ForestDiffusionGenerator(BaseGenerator):
|
|
|
29
29
|
trained XGBoost models are kept in memory. Default: None.
|
|
30
30
|
num_timesteps: The number of noise levels to use. Default: 50.
|
|
31
31
|
diffusion_type: The type of diffusion to use. Default: "vp".
|
|
32
|
-
multi_output: Whether to use multi-output XGBoost
|
|
32
|
+
multi_output: Whether to use multi-output XGBoost. Default: False.
|
|
33
33
|
xgb_hypers: The hyperparameters for the XGBoost models. Default: {"n_estimators": 100, "max_depth": 6, "early_stopping_rounds": 50}.
|
|
34
34
|
noise_samples_per_row: The number of different noise samples per real data sample. Default: 100.
|
|
35
35
|
eps: The timestep to stop generation at. Default: 0.001.
|
|
36
36
|
beta_min: The minimum beta for the diffusion. Default: 0.1.
|
|
37
37
|
beta_max: The maximum beta for the diffusion. Default: 8.
|
|
38
38
|
n_jobs: The number of parallel jobs to use. Default: -1.
|
|
39
|
-
backend: The backend to use. Default: "loky".
|
|
40
|
-
n_batch: The number of batches to use. Default: -1.
|
|
39
|
+
backend: The joblib backend to use. Default: "loky".
|
|
40
|
+
n_batch: The number of batches to use in XGBoost data iterator. Set to <=0 to disable data iterator. Default: -1.
|
|
41
41
|
|
|
42
42
|
Example:
|
|
43
43
|
>>> import pandas as pd
|