Skip to content

Has this approach been used / tested with the new WAN 2.1 diffusers pipeline? #10

Open
@ukaprch

Description

@ukaprch

I ran your approach with the diffusers library pipeline for WAN T2V 14 B and received the following messages:

Warning: DiT library is not accessible. DiTCacheHelper cannot be used.
Warning: CalibrationHelper not imported.
Warning: DiffuserCalibrationHelper not imported. Ensure Diffusers is installed.

bypassing smoothcache processing. I used the code below:

if device == 'cpu': pipe.enable_model_cpu_offload() else: pipe = pipe.to(device=device) # Import SmoothCacheHelper from SmoothCache import DiffuserCacheHelper import json # Initialize the DiffuserCacheHelper with the model with open("./SmoothCache/smoothcache_schedules/diffuser_schedule.json", "r") as f: schedule = json.load(f) cache_helper = DiffuserCacheHelper(pipe.transformer, schedule=schedule) # Enable the caching helper cache_helper.enable()
Is there an alternative approach that can be used?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions