`bridge.models.gemma_vl.data.collate_fn`#

Gemma VL collator implementations.

Module Contents#

`gemma3_vl_collate_fn`	Collate function for Gemma3 VL models.
`gemma4_vl_collate_fn`	Collate function for Gemma4 VL models.

bridge.models.gemma_vl.data.collate_fn.gemma3_vl_collate_fn( examples: list, processor, *, visual_keys: collections.abc.Sequence[str] = THW_GRID_VISUAL_KEYS, min_pixels: int | None = None, max_pixels: int | None = None, sequence_length: int | None = None, pad_to_max_length: bool = False, pad_to_multiple_of: int = 128, enable_in_batch_packing: bool = False, in_batch_packing_pad_to_multiple_of: int = 1, ) → dict[str, torch.Tensor]#: Collate function for Gemma3 VL models.

bridge.models.gemma_vl.data.collate_fn.gemma4_vl_collate_fn( examples: list, processor, *, visual_keys: object = None, min_pixels: int | None = None, max_pixels: int | None = None, sequence_length: int | None = None, pad_to_max_length: bool = False, pad_to_multiple_of: int = 128, enable_in_batch_packing: bool = False, in_batch_packing_pad_to_multiple_of: int = 1, ) → dict[str, torch.Tensor]#: Collate function for Gemma4 VL models.