bridge.models.qwen_vl.qwen_vl_provider

Module Contents

Classes

Qwen25VLModelProvider

Base model provider for Qwen 2.5 VL Models.

API

class bridge.models.qwen_vl.qwen_vl_provider.Qwen25VLModelProvider

Bases: megatron.bridge.models.Qwen2ModelProvider

Base model provider for Qwen 2.5 VL Models.

scatter_embedding_sequence_parallel: bool

False

position_embedding_type: str

‘mrope’

mrope_section: List[int]

‘field(…)’

vision_config: transformers.models.qwen2_5_vl.modeling_qwen2_5_vl.Qwen2_5_VLVisionConfig

‘field(…)’

bos_token_id: int

151643

eos_token_id: int

151645

vision_start_token_id: int

151652

vision_end_token_id: int

151653

vision_token_id: int

151654

image_token_id: int

151655

video_token_id: int

151656

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,

) → bridge.models.qwen_vl.modeling_qwen25_vl.Qwen25VLModel

provide_language_model(

pre_process=None,

post_process=None,

vp_stage=None,

) → megatron.core.models.gpt.GPTModel