Trending storyDeveloping

ModelScope adds OPD few-step diffusion training workflow to DiffSynth-Studio

1 article1 sourcesince Oct 15Last article 54m ago · Oct 15, 6:00 AM ET

Overview

AISummary of 1 article

ModelScope has added an OPD training workflow to DiffSynth-Studio for training few-step diffusion models, released under the Apache 2.0 license.

According to ModelScope's post on X, training typically converges within 2 GPU-hours. The company reports distilling FLUX.2-klein-base-4B and Qwen-Image-2.1 from 50 to 8 steps, and Qwen-Image-2.1-Turbo from 8 to 3 steps, while keeping generation quality. The workflow trains on student-generated trajectories with teacher guidance. These results are ModelScope's own claims and have not been independently verified.

Written by AI from the articles below · updated Oct 15, 6:14 AM ET

Check the sources:

Latest developmentOct 15, 6:00 AM ET

ModelScope adds OPD training workflow to DiffSynth-Studio for few-step diffusion models

Article timeline

The articles in this story. Times are ET.

Oct 15

6:00 AM
ModelScope@ModelScope2022Official
ModelScope adds OPD training workflow to DiffSynth-Studio for few-step diffusion models
AIModelScope has added an OPD training workflow to DiffSynth-Studio for training few-step diffusion models. It reports training convergence typically within 2 GPU-hours and distillation of FLUX.2-klein-base-4B and Qwen-Image-2.1 from 50 to 8 steps, and Qwen-Image-2.1-Turbo from 8 to 3 steps, while keeping generation quality. The workflow trains on student-generated trajectories with teacher guidance, and the code is released under the Apache 2.0 license.

Heat trend

Not enough continuous observations to show a trend yet.