bridge.models.gemma_vl.gemma4_vl_provider

Gemma 4 VL model provider.

Module Contents

Classes

Gemma4VLModelProvider

Model provider for Gemma 4 Vision-Language models.

API

class bridge.models.gemma_vl.gemma4_vl_provider.Gemma4VLModelProvider

Bases: megatron.bridge.models.gemma.gemma4_provider.Gemma4ModelProvider

Model provider for Gemma 4 Vision-Language models.

Extends Gemma4ModelProvider with vision tower config, multimodal projector config, and token IDs for vision-text fusion.

scatter_embedding_sequence_parallel: bool

False

vision_config: Any

None

text_config: Any

None

vision_soft_tokens_per_image: int

280

bos_token_id: int

2

eos_token_id: int

1

image_token_id: int

258880

video_token_id: int

258884

freeze_language_model: bool

False

freeze_vision_model: bool

False

freeze_vision_projection: bool

False

provide(

pre_process=None,

post_process=None,

vp_stage=None,

) → megatron.bridge.models.gemma_vl.modeling_gemma4_vl.Gemma4VLModel

provide_language_model(

pre_process=None,

post_process=None,

vp_stage=None,

) → megatron.core.models.gpt.GPTModel