Can DALI be integrated into HuggingFace Trainer? about dali HOT 9 OPEN

Hi @ShyFoo,

Thank you for reaching out. While we haven't tried this directly yet DALI provides compatibility with PyTorch so you can create a DALI iterator that would behave as PyTorch one.
Have you tried this yet? Have you encountered any issues?

from dali.

I just checked and unfortunately, HuggingFace trainer (I checked only one example so far) expects PyTorch DataIterator (checks for the type of passes object). I still believe it is possible to make DALI work there but I may need a bit of adjustments to the internal methods. If you have some spare time we would be more than happy to review a PR with such an example.

from dali.

I just checked and unfortunately, HuggingFace trainer (I checked only one example so far) expects PyTorch DataIterator (checks for the type of passes object). I still believe it is possible to make DALI work there but I may need a bit of adjustments to the internal methods. If you have some spare time we would be more than happy to review a PR with such an example.

Thanks for your reply. Yeah, I have tried integrating DALI into the HuggingFace trainer. As you mentioned, it seems to expect either torch.utils.data.Dataset or torch.utils.data.IterableDataset as input. It might be possible to customize a data collator, which can preprocess data in a DALI pipeline, for the HuggingFace trainer, but I'm not sure if it will work. If you have any idea for this, please let me know. I'm sure it would be a huge step for the whole community!

from dali.

I did some basic investigation:

• for CV tasks DALI needs to have such wrapper:

  class DALIwrapper(DALIGenericIterator):
      def __init__(self, *args, **kargs):
          super().__init__(*args, **kargs)
  
      def __next__(self):
          batch = super().__next__()
          return batch[0]
      
  pipe_train = GetDaliPipeline(...)
  data_train = DALIwrapper([pipe_train], ["pixel_values", "labels"], reader_name="Reader")
  

• trainer needs to have these methods replaced:

  trainer.get_train_dataloader = lambda : data_train
  trainer.get_eval_dataloader = lambda x: x
  

  so DALI dataloader is used as is without wrapping into PyTorch DataIterator
• there is probably another layer of complexity using DALI with a distributed strategy which is handled by accelerate

from dali.

For the third point, get_train_dataloader returns self.accelerator.prepare(DataLoader(train_dataset, **dataloader_params)) for a distributed training or evaluation. So what should I do in this setting? Hope for your help.

from dali.

@ShyFoo - I'm afraid I don't know the answer. My idea was to avoid wrapping DALI into DataLoader as it serves as one already. If you have time I would be more than happy to learn your findings.

from dali.

e basic investigation:

Got it. I have tried using DALI + DeepSpeed, where I have to define a triaing loop and other practical functions myself, instead of using the Huggingface Trainer for convenience, so I prefer to use DALI, DeepSpeed, and Huggingface Trainer simultaneously if possible.

from dali.

If you have any examples of DALI + DeepSpeed feel free to post them as PR to DALI - we would be more than happy to enrich our documentation and the base of samples.

from dali.

Well, I'd be happy to.

from dali.