hristo-vrigazov/mmap.ninja

Issues

• Parallel batch collection is significantly slower

  #18 opened 3 months ago by jzluo
  1

• Advanced dataloading

  #16 opened 6 months ago by mileseverett
  2

• Access to Collab Notebooks

  #14 opened 9 months ago by rfalcon100
  3

• copying with `wrapper_fn`

  #13 opened a year ago by ahalev
  3

• Compatibility with DataLoaders and Multi-GPU on DDP

  #9 opened a year ago by Yusepp
  2

• Unused parameter n in numpy.from_generator

  #12 opened a year ago by SimonKoop
  2

• Trouble opening a StringsMmap read-only

  #10 opened 2 years ago by ndandanov
  1

• Getting slower and slower

  #8 opened 2 years ago by nctamer
  8

• Segmentation fault when integrating with pyarrow batches

  #7 opened 2 years ago by rodrigolece
  4

• How does it compare to pyarrow, hdf5 or lmdb?

  #6 opened 2 years ago by TheShadow29
  1

• Just a question

  #4 opened 2 years ago by slerman12
  6

• Is it working for COCO train2017?

  #3 opened 2 years ago by hanjoo-cho
  4

• Suggest adding parents=True to Path.mkdir()

  #1 opened 2 years ago by versatran01
  3