fastvideo.pipelines

fastvideo.pipelines#

Diffusion pipelines for fastvideo.

This package contains diffusion pipelines for generating videos and images.

Subpackages#

Submodules#

Package Contents#

Classes#

PipelineWithLoRA

Type for a pipeline that has both ComposedPipelineBase and LoRAPipeline functionality.

Functions#

build_pipeline

Only works with valid hf diffusers configs. (model_index.json) We want to build a pipeline based on the inference args mode_path:

Data#

logger

API#

class fastvideo.pipelines.PipelineWithLoRA(*args, **kwargs)[source]#

Bases: fastvideo.pipelines.lora_pipeline.LoRAPipeline, fastvideo.pipelines.composed_pipeline_base.ComposedPipelineBase

Type for a pipeline that has both ComposedPipelineBase and LoRAPipeline functionality.

Initialization

Initialize the pipeline. After init, the pipeline should be ready to use. The pipeline should be stateless and not hold any batch state.

fastvideo.pipelines.build_pipeline(fastvideo_args: fastvideo.fastvideo_args.FastVideoArgs, pipeline_type: fastvideo.pipelines.pipeline_registry.PipelineType | str = PipelineType.BASIC) → fastvideo.pipelines.PipelineWithLoRA[source]#

Only works with valid hf diffusers configs. (model_index.json) We want to build a pipeline based on the inference args mode_path:

  1. download the model from the hub if it’s not already downloaded

  2. verify the model config and directory

  3. based on the config, determine the pipeline class

fastvideo.pipelines.logger[source]#

‘init_logger(…)’

Contents