Skip to content

vllm_omni.model_extras.video_generation ¶

VideoGenerationDefaults `dataclass` ¶

Model-owned defaults consumed by shared video examples.

default_negative_prompt `class-attribute` `instance-attribute` ¶

default_negative_prompt: str | None = ''

dimension_multiple `class-attribute` `instance-attribute` ¶

dimension_multiple: int = 16

duration_seconds `class-attribute` `instance-attribute` ¶

duration_seconds: float | None = None

flow_shift `class-attribute` `instance-attribute` ¶

flow_shift: float | None = None

fps `instance-attribute` ¶

fps: float

guidance_scale `class-attribute` `instance-attribute` ¶

guidance_scale: float | None = None

height `instance-attribute` ¶

height: int

max_area `property` ¶

max_area: int

num_frames `instance-attribute` ¶

num_frames: int

num_inference_steps `instance-attribute` ¶

num_inference_steps: int

output `class-attribute` `instance-attribute` ¶

output: str | None = None

width `instance-attribute` ¶

width: int

cli_defaults ¶

cli_defaults() -> dict[str, Any]