patch_save_to_disk
#340
by
SidMahesh
- opened
- geneformer/classifier.py +8 -8
geneformer/classifier.py
CHANGED
@@ -445,8 +445,8 @@ class Classifier:
|
|
445 |
test_data_output_path = (
|
446 |
Path(output_directory) / f"{output_prefix}_labeled_test"
|
447 |
).with_suffix(".dataset")
|
448 |
-
data_dict["train"].save_to_disk(train_data_output_path)
|
449 |
-
data_dict["test"].save_to_disk(test_data_output_path)
|
450 |
elif (test_size is not None) and (self.classifier == "cell"):
|
451 |
if 1 > test_size > 0:
|
452 |
if attr_to_split is None:
|
@@ -461,8 +461,8 @@ class Classifier:
|
|
461 |
test_data_output_path = (
|
462 |
Path(output_directory) / f"{output_prefix}_labeled_test"
|
463 |
).with_suffix(".dataset")
|
464 |
-
data_dict["train"].save_to_disk(train_data_output_path)
|
465 |
-
data_dict["test"].save_to_disk(test_data_output_path)
|
466 |
else:
|
467 |
data_dict, balance_df = cu.balance_attr_splits(
|
468 |
data,
|
@@ -483,19 +483,19 @@ class Classifier:
|
|
483 |
test_data_output_path = (
|
484 |
Path(output_directory) / f"{output_prefix}_labeled_test"
|
485 |
).with_suffix(".dataset")
|
486 |
-
data_dict["train"].save_to_disk(train_data_output_path)
|
487 |
-
data_dict["test"].save_to_disk(test_data_output_path)
|
488 |
else:
|
489 |
data_output_path = (
|
490 |
Path(output_directory) / f"{output_prefix}_labeled"
|
491 |
).with_suffix(".dataset")
|
492 |
-
data.save_to_disk(data_output_path)
|
493 |
print(data_output_path)
|
494 |
else:
|
495 |
data_output_path = (
|
496 |
Path(output_directory) / f"{output_prefix}_labeled"
|
497 |
).with_suffix(".dataset")
|
498 |
-
data.save_to_disk(data_output_path)
|
499 |
|
500 |
def train_all_data(
|
501 |
self,
|
|
|
445 |
test_data_output_path = (
|
446 |
Path(output_directory) / f"{output_prefix}_labeled_test"
|
447 |
).with_suffix(".dataset")
|
448 |
+
data_dict["train"].save_to_disk(str(train_data_output_path))
|
449 |
+
data_dict["test"].save_to_disk(str(test_data_output_path))
|
450 |
elif (test_size is not None) and (self.classifier == "cell"):
|
451 |
if 1 > test_size > 0:
|
452 |
if attr_to_split is None:
|
|
|
461 |
test_data_output_path = (
|
462 |
Path(output_directory) / f"{output_prefix}_labeled_test"
|
463 |
).with_suffix(".dataset")
|
464 |
+
data_dict["train"].save_to_disk(str(train_data_output_path))
|
465 |
+
data_dict["test"].save_to_disk(str(test_data_output_path))
|
466 |
else:
|
467 |
data_dict, balance_df = cu.balance_attr_splits(
|
468 |
data,
|
|
|
483 |
test_data_output_path = (
|
484 |
Path(output_directory) / f"{output_prefix}_labeled_test"
|
485 |
).with_suffix(".dataset")
|
486 |
+
data_dict["train"].save_to_disk(str(train_data_output_path))
|
487 |
+
data_dict["test"].save_to_disk(str(test_data_output_path))
|
488 |
else:
|
489 |
data_output_path = (
|
490 |
Path(output_directory) / f"{output_prefix}_labeled"
|
491 |
).with_suffix(".dataset")
|
492 |
+
data.save_to_disk(str(data_output_path))
|
493 |
print(data_output_path)
|
494 |
else:
|
495 |
data_output_path = (
|
496 |
Path(output_directory) / f"{output_prefix}_labeled"
|
497 |
).with_suffix(".dataset")
|
498 |
+
data.save_to_disk(str(data_output_path))
|
499 |
|
500 |
def train_all_data(
|
501 |
self,
|