Hugging Face's logo

Datasets:
HuggingFaceM4
/
pmd

Dataset card Files
xet
Community
2
thomasw21 commited on
Commit
44f5a1f
·
1 Parent(s): 20f4f97

Get PMD running

Browse files
Files changed (1) hide show
  1. pmd.py +65 -68
pmd.py CHANGED
@@ -57,7 +57,6 @@ _FEATURES = datasets.Features(
57
  }
58
  )
59
 
60
-
61
  def json_serializer(o):
62
  if isinstance(o, datetime):
63
  return str(o)
@@ -87,7 +86,6 @@ class DatasetsLoader(BaseLoader):
87
  num_proc: int,
88
  datasets_batch_size: int = 1000,
89
  ):
90
- # TODO @thomasw21 do we somehow want to de-correlate both writer and datasets batch_size
91
  super(DatasetsLoader, self).__init__(
92
  source=dataset_name, split=split, writer_batch_size=datasets_batch_size
93
  )
@@ -936,7 +934,6 @@ class PMDConfig(datasets.BuilderConfig):
936
  self.writer_batch_size = writer_batch_size
937
 
938
 
939
- # TODO @thomasw21 ArrowBasedBuilder to be able to return batches
940
  class PMD(datasets.ArrowBasedBuilder):
941
  """Builder for Open Images subset of PMD."""
942
 
@@ -957,71 +954,71 @@ class PMD(datasets.ArrowBasedBuilder):
957
  name=split_name,
958
  gen_kwargs={
959
  "loaders": [
960
- # COCOloader(
961
- # dl_manager=dl_manager,
962
- # split=split_name,
963
- # num_proc=self.config.num_proc,
964
- # chunk_size=self.config.chunk_size,
965
- # writer_batch_size=self.config.writer_batch_size,
966
- # ),
967
- # SBULoader(
968
- # split=split_name,
969
- # datasets_batch_size=self.config.datasets_batch_size,
970
- # num_proc=self.config.num_proc,
971
- # ),
972
- # LocalizedNarrativesOpenImagesLoader(
973
- # dl_manager=dl_manager,
974
- # split=split_name,
975
- # num_proc=self.config.num_proc,
976
- # chunk_size=self.config.chunk_size,
977
- # writer_batch_size=self.config.writer_batch_size,
978
- # ),
979
- # LocalizedNarrativesCOCOLoader(
980
- # dl_manager=dl_manager,
981
- # split=split_name,
982
- # num_proc=self.config.num_proc,
983
- # chunk_size=self.config.chunk_size,
984
- # writer_batch_size=self.config.writer_batch_size,
985
- # ),
986
- # LocalizedNarrativesFlickr30kLoader(
987
- # dl_manager=dl_manager,
988
- # split=split_name,
989
- # num_proc=self.config.num_proc,
990
- # chunk_size=self.config.chunk_size,
991
- # writer_batch_size=self.config.writer_batch_size,
992
- # ),
993
- # LocalizedNarrativesADE20kLoader(
994
- # dl_manager=dl_manager,
995
- # split=split_name,
996
- # num_proc=self.config.num_proc,
997
- # chunk_size=self.config.chunk_size,
998
- # writer_batch_size=self.config.writer_batch_size,
999
- # ),
1000
- # ConceptualCaptions(
1001
- # split=split_name,
1002
- # num_proc=self.config.num_proc,
1003
- # datasets_batch_size=self.config.datasets_batch_size,
1004
- # ),
1005
- # VisualGenomeLoader(
1006
- # split=split_name,
1007
- # num_proc=self.config.num_proc,
1008
- # datasets_batch_size=self.config.datasets_batch_size,
1009
- # ),
1010
- # WITLoader(
1011
- # split=split_name,
1012
- # num_proc=self.config.num_proc,
1013
- # datasets_batch_size=self.config.datasets_batch_size,
1014
- # ),
1015
- # Conceptual12MLoader(
1016
- # split=split_name,
1017
- # num_proc=self.config.num_proc,
1018
- # datasets_batch_size=self.config.datasets_batch_size,
1019
- # ),
1020
- # RedCapsLoader(
1021
- # split=split_name,
1022
- # num_proc=self.config.num_proc,
1023
- # datasets_batch_size=self.config.datasets_batch_size,
1024
- # ),
1025
  YFCC100MLoader(
1026
  dl_manager=dl_manager,
1027
  split=split_name,
 
57
  }
58
  )
59
 
 
60
  def json_serializer(o):
61
  if isinstance(o, datetime):
62
  return str(o)
 
86
  num_proc: int,
87
  datasets_batch_size: int = 1000,
88
  ):
 
89
  super(DatasetsLoader, self).__init__(
90
  source=dataset_name, split=split, writer_batch_size=datasets_batch_size
91
  )
 
934
  self.writer_batch_size = writer_batch_size
935
 
936
 
 
937
  class PMD(datasets.ArrowBasedBuilder):
938
  """Builder for Open Images subset of PMD."""
939
 
 
954
  name=split_name,
955
  gen_kwargs={
956
  "loaders": [
957
+ COCOloader(
958
+ dl_manager=dl_manager,
959
+ split=split_name,
960
+ num_proc=self.config.num_proc,
961
+ chunk_size=self.config.chunk_size,
962
+ writer_batch_size=self.config.writer_batch_size,
963
+ ),
964
+ SBULoader(
965
+ split=split_name,
966
+ datasets_batch_size=self.config.datasets_batch_size,
967
+ num_proc=self.config.num_proc,
968
+ ),
969
+ LocalizedNarrativesOpenImagesLoader(
970
+ dl_manager=dl_manager,
971
+ split=split_name,
972
+ num_proc=self.config.num_proc,
973
+ chunk_size=self.config.chunk_size,
974
+ writer_batch_size=self.config.writer_batch_size,
975
+ ),
976
+ LocalizedNarrativesCOCOLoader(
977
+ dl_manager=dl_manager,
978
+ split=split_name,
979
+ num_proc=self.config.num_proc,
980
+ chunk_size=self.config.chunk_size,
981
+ writer_batch_size=self.config.writer_batch_size,
982
+ ),
983
+ LocalizedNarrativesFlickr30kLoader(
984
+ dl_manager=dl_manager,
985
+ split=split_name,
986
+ num_proc=self.config.num_proc,
987
+ chunk_size=self.config.chunk_size,
988
+ writer_batch_size=self.config.writer_batch_size,
989
+ ),
990
+ LocalizedNarrativesADE20kLoader(
991
+ dl_manager=dl_manager,
992
+ split=split_name,
993
+ num_proc=self.config.num_proc,
994
+ chunk_size=self.config.chunk_size,
995
+ writer_batch_size=self.config.writer_batch_size,
996
+ ),
997
+ ConceptualCaptions(
998
+ split=split_name,
999
+ num_proc=self.config.num_proc,
1000
+ datasets_batch_size=self.config.datasets_batch_size,
1001
+ ),
1002
+ VisualGenomeLoader(
1003
+ split=split_name,
1004
+ num_proc=self.config.num_proc,
1005
+ datasets_batch_size=self.config.datasets_batch_size,
1006
+ ),
1007
+ WITLoader(
1008
+ split=split_name,
1009
+ num_proc=self.config.num_proc,
1010
+ datasets_batch_size=self.config.datasets_batch_size,
1011
+ ),
1012
+ Conceptual12MLoader(
1013
+ split=split_name,
1014
+ num_proc=self.config.num_proc,
1015
+ datasets_batch_size=self.config.datasets_batch_size,
1016
+ ),
1017
+ RedCapsLoader(
1018
+ split=split_name,
1019
+ num_proc=self.config.num_proc,
1020
+ datasets_batch_size=self.config.datasets_batch_size,
1021
+ ),
1022
  YFCC100MLoader(
1023
  dl_manager=dl_manager,
1024
  split=split_name,