Issue in TransDataCollatorForLanguageModeling

[JulinaM]

AttributeError: 'TransDataCollatorForLanguageModeling' object has no attribute '_tensorize_batch'

[ianbenlolo]

I replaced that line with

if isinstance(examples[0], (list, tuple, np.ndarray)):
    examples = [torch.tensor(e, dtype=torch.long) for e in examples]

and it does what (i think) is supposed to happen

[amahonycxnpl]

were either of you guys able to successfully get evaluation metrics on the TabBERT model? Struggling with this.