Map method doesn't work with custom dataset

I have created custom dataset with __get__ and __len__ functions, but using map method on the instance of the datasets returns me an error, that dataset has no attribtute map

Here is dataset:

class trainingDataset(Dataset):
  def __init__(self):
    super().__init__()
    self.trainingAudio = audioWorker()
    self.trainingAudio.set = "training"
  def __len__(self):
    return self.trainingAudio.vocabLen
    
  def __getitem__(self, idx):
    return self.trainingAudio.getAudio(idx)

and here is what I am trying to do:

def prepare_dataset(batch):
  audio = batch["audio"]
  # batched output is "un-batched"
  batch["input_values"] = processor(audio["array"], sampling_rate=audio["sampling_rate"]).input_values[0]
  batch["input_length"] = len(batch["input_values"])
  
  with processor.as_target_processor():
      batch["labels"] = processor(batch["sentence"]).input_ids
  return batch

testDataset = testDataset.map(prepare_dataset)

AttributeError: 'testDataset' object has no attribute 'map'

It’s looking really strange, because it was working well before :confused:

That’s strange, as I don’t think torch.utils.data.Dataset ever provided a map method. Also the current implementation in the master branch doesn’t support it as seen here.
Were you maybe using the new torchdata.datapipes.map.Mapper functionality?