Support `BatchFeature` in `LengthGroupedSampler` for Multimodal compatibility

[npurson]

Feature request

I am currently fine-tuning a multimodal model (Qwen2.5-VL) using the official Trainer. The training fails during the dataset length inference step in LengthGroupedSampler because the code strictly checks for dict or BatchEncoding, but multimodal processors often return BatchFeature.

Specifically, the following check raises a ValueError:

transformers/src/transformers/trainer_pt_utils.py

Line 471 in 471d7ce

if not isinstance(dataset[0], (dict, BatchEncoding)) or model_input_name not in dataset[0]:

transformers/src/transformers/trainer_pt_utils.py

Line 531 in 471d7ce

if not isinstance(dataset[0], (dict, BatchEncoding)) or model_input_name not in dataset[0]:

Motivation

As above.

Your contribution

Simply adding BatchFeature in the type check.

[zucchini-nlp]

@npurson sounds good, do you want to submit a PR?